Hydrogen sulfide and inflammation: the good, the bad, the ugly and the promising

Abstract

Hydrogen sulfide is rapidly gaining ground as a physiological mediator of inflammation, but there is no clear consensus as to its precise role in inflammatory signaling. This article discusses the disparate anti-inflammatory (‘the good’) and proinflammatory (‘the bad’) effects of endogenous and pharmacological H₂S in disparate animal model and cell culture systems. We also discuss ‘the ugly’, such as problems of using wholly specific inhibitors of enzymatic H₂S synthesis, and the use of pharmacological donor compounds, which release H₂S too quickly to be physiologically representative of endogenous H₂S synthesis. Furthermore, recently developed slow-release H₂S donors, which offer a more physiological approach to understanding the complex role of H₂S in acute and chronic inflammation (‘the promising’) are discussed.

Keywords::

• β-cyanoalanine
• aminooxyacetate
• arthritis
• asthma
• chronic obstructive pulmonary disorder
• colitis
• cystathionine β-synthase
• cystathionine γ-lyase
• GYY4137
• NSAID
• pain
• pancreatitis
• propargylglycine
• shock

Figure 1. Comparison of synovial fluid concentrations of H₂S in inflammatory and noninflammatory arthritides.

Plasma and synovial fluid aspirates were obtained from healthy controls, patients with OA (n = 5), RA (n = 27), ReA (n = 7) and PsA (n = 5) and H₂S levels determined by zinc-trap spectrophotometry, as described in [18,22]. All patients and volunteers in this study were enrolled following institutional ethical approval (North Devon and Exeter Research Ethics Committee #04/Q2102/87 and #05/Q2103/91). All patients with RA and OA fulfilled the American College of Rheumatology criteria [143] and attended the department of Rheumatology, Royal Devon and Exeter Hospital Trust and had given informed written consent. Data are shown as mean ± standard deviation. Statistical analysis was by ANOVA.

H₂S: Hydrogen sulfide; OA: Osteoarthritis; PsA: Psoriatic arthritis; RA: Rheumatoid arthritis; ReA: Reactive arthritis.

Figure 2. Chemical structures of commonly used inhibitors of CSE and CBS.

Inhibitors of CSE: dL-propargylglycine and β-cyanoalanine. Inhibitor of CBS: aminooxyacetate.

CBS: Cystathionine β-synthase; CSE: Cystathionine γ-lyase.

Figure 3. Chemical structures of some fast- and slow-releasing H₂S donors.

ADT-OH: 5-(4-hydroxyphenyl)-3H-1,2-dithiole-3-thione; H₂S: Hydrogen sulfide.

Inflammation

Inflammation is a protective physiological reaction of vascularized tissue to local injury or tissue destruction [1–3]. In its broadest sense, inflammation is the host response to tissue injury to remove injurious stimuli and initiate a healing process. It is a complex network of coordinated cellular responses designed to destroy, dilute or ‘wall-off’ both the noxious stimuli and the injured tissue [3,4]. The five clinical (‘cardinal’) signs of inflammation are redness (rubor), heat (calor), swelling (tumor), pain (dolor) and eventually loss of function (functio laesa) [2,3]. These cardinal signs represent the macroscopic culmination of a multitude of molecular and cellular processes, many of which have become well defined over the last century, and may be reproduced in convenient in vitro experimental systems or in more complex and, perhaps more importantly, in relevant animal models in vivo. Although detailed descriptions of the myriad inflammatory pathways are the subject of a plethora of review articles and medical text books, for the context of this current article, a brief discussion of the general inflammatory process is pertinent.

The inflammatory response is often categorized according to the duration and kinetics of the reaction and can be classified as either acute or chronic [2,3]. In general terms, acute inflammation is the immediate and early host response to an injurous agent. It is mediated by an increase in blood flow into injured tissues primarily from arteriolar dilatation and the opening of capillary beds. There is also increased movement of leukocytes (in particular granulocytes, such as neutrophils) and increased leukocyte adhesion to the vascular endothelium (via adhesion molecules). Leukocytes phagocytose the injurious agent and release toxic metabolites and proteases, which potentially causes tissue damage. These processes result in a cascade of biochemical mediators, such as vasoactive amines (e.g., histamine and 5-hydroxytryptamine), plasma proteases (e.g., the complement, kinin and clotting systems) and arachidonic acid metabolites (e.g., prostaglandins, leukotrienes and lipoxins) [5]. A complex array of cytokines and chemokines are also produced by inflammatory cells and the vascular endothelium, which regulate lymphocyte function through activating (e.g., IL-2, IL-4, IFN-α and IFN-β) or inhibiting (e.g., IL-10 and TGF-β) immune responses. Cytokines activate inflammatory cells (e.g., TNF-α, IL-1β, IFN-γ and IL-6) or stimulate hematopoiesis and leukocyte growth and differentiation (e.g., IL-3 and IL-7) [5]. Physiological gaseous mediators, such as nitric oxide (^•NO) and carbon monoxide (CO; reviewed in explicit detail elsewhere [6–10]) have been proposed to induce, inhibit and regulate the inflammatory process.

More recently, a third endogenous gas, hydrogen sulfide (H₂S), has been proposed as an additional inflammatory mediator [11], although the precise role for H₂S is controversial (see later; detailed discussion on endogenous synthesis of H₂S is contained elsewhere in this article). Therefore, acute inflammation manifests as vascular changes, edema and largely neutrophillic infiltration. Examples of acute inflammation are septic shock, multiple organ failure and pancreatitis.

Chronic inflammation, on the other hand, is difficult to define precisely. In general terms, it is inflammation of prolonged duration (weeks or months), in which active inflammation, tissue destruction and attempts at repair proceed simultaneously [2], and leads to several distinct processes:

• • A progressive shift in the types of cells present at the site of inflammation (e.g., from granulocytes to mononuclear cells, such as macrophages, lymphocytes and plasma cells, which characterize the persistent reaction to injury);

• • Tissue destruction mediated predominantly by the inflammatory cells;

• • Connective tissue replacement of damaged tissue via microvascular proliferation (angiogenesis) and fibrosis (attempts at healing).

Although chronic inflammation may follow acute inflammation, frequently, it has insidious beginnings as a simmering low-grade and often asymptomatic response, so-called ‘subclinical inflammation’. Types of chronic inflammation include some of the most common and debilitating human diseases, such as rheumatoid arthritis, tuberculosis, asthma, inflammatory bowel disease, vasculitis and Crohn’s disease. Consequently, pharmacological agents to treat these disorders represent a multibillion dollar industry and novel chemical entities, which prevent, inhibit or resolve inflammation, are of continual interest to the global pharmaceutical industry.

Recent evidence for a role of H₂S in inflammation

H₂S synthesis & biochemistry

The synthesis pathways and biochemistry of H₂S are reviewed extensively throughout this article. However, a generalized overview to familarize the reader with the topic is pertinent. H₂S has been proposed as a third physiological gaseous mediator along with CO and ^•NO, for which there appears to be considerable interplay (reviewed elsewhere [6,12]). Its formation in vivo is catalyzed predominantly by the pyridoxal phosphate-dependent enzymes cystathionine-γ-lyase (CSE) and cystathionine β-synthase (CBS), utilizing the amino acids l-cysteine, l-homocysteine and l-cystathionine (reviewed in [12]). In the last decade, H₂S has been shown to regulate a steadily growing list of physiological functions, such as blood pressure [13], insulin secretion [14,15], nociception [16], learning and memory [17] and inflammation [6], in part via the modulation of K_ATP channel activity. As such, modulation of endogenous H₂S through either inhibitors of its enzymatic synthesis or H₂S-donor molecules represent viable and potentially therapeutic targets for a variety of human diseases where perturbed H₂S synthesis is involved, such as hypertension [13,18,19], diabetes and obesity [18], hemorrhagic [20], endotoxic shock [21] and chronic inflammation (e.g., arthritis [22] and inflammatory bowel disease [23].

However, as we will discuss, the precise role for H₂S is controversial, and as with any new and emerging field of research, progress has been hampered by the limitations on selective tools, such as CSE and CBS inhibitors and H₂S-donor molecules that model endogenous enzymatic H₂S synthesis. It should be noted that H₂S is a weak acid (pKa = 6.96), and at physiological pH 7.4, H₂S will always exist as approximately a third H₂S and two-thirds H⁺ and HS^- (hydrosulfide anion). At present, it is unknown whether the biological effects of H₂S are mediated directly by H₂S or by HS^-, or whether a combination of both species is required. During the inflammatory process, mildly acidic conditions may predominate, which would shift the equilibrium towards H₂S. Therefore, we prudently use the term ‘H₂S’ to encompass H₂S and HS^-.

H₂S & acute whole-body inflammation (sepsis)

Sepsis and its sequelae (e.g., septic shock and multiple organ failure) are the most common cause of death in medical and surgical intensive units [24]. The predominant view of sepsis is one of acute whole-body inflammation in the presence of a known or suspected infection, most commonly caused by bacteria, but also induced by fungi, viruses and protozoa [2,24], resulting in massively elevated levels of inflammatory markers (e.g., acute-phase proteins, cytokines and NO and prostanoids). This results in tissue hypoperfusion and organ dysfunction (severe sepsis) and, in septic shock, persistent hypotension [2]. Both hypoperfusion and hypotension damage the vascular endothelium and impair tissue respiration leading to high mortality rates – 16% in sepsis to up to 60% in patients with septic shock [25].

Bacterial endotoxins, such as lipopolysaccharide (LPS), have previously been shown to induce excessive activation and upregulation of vascular K_ATP channels, to induce hypotension and to substantially reduce vascular sensitivity to vasoconstrictive agents [26,27]. H₂S has been proposed as a potential endogenous ligand for K_ATP channels and induces K_ATP channel-mediated vasorelaxation [28] in several vascular tissues, promoting speculation that H₂S may play a role in endotoxic shock, an important form of acute inflammation, as mentioned. Indeed, early studies using cecal ligation and puncture (CLP) or injection of bacterial LPS into rats or mice significantly increased the expression of CSE and CBS and H₂S synthesis in plasma, vascular tissues, lung, liver, kidney and pancreas [21,29–32]. This suggested that, as with ^•NO synthesis (via inducible nitric oxide synthase [iNOS]), H₂S synthesis is also an inducible phenomenon. Plasma H₂S levels in patients with septic shock were also reported to be elevated up to four-times compared with control subjects (up to 200 µmol/l in one patient) [21]. At these plasma concentrations, H₂S induces K_ATP-dependent vasodilatation in anesthetized animals and in isolated aortic rings [28,33,34]. Conversely, administration of the CSE inhibitor propargylglycine (PAG; see later) to LPS- or CLP-treated animals decreased plasma H₂S levels significantly, tissue edema and increased survival [21,29,30], suggesting that elevated plasma H₂S was likely to be associated with the hemodynamic collapse observed in septic shock [21]. It should be noted that the protective effects of PAG may be independent of the improvement of hemodynamics (or CSE), since PAG treatment alone did not alter systemic blood pressure or affect LPS-induced hypotension but did significantly reduce the hepatic, pancreatic and neuromuscular damage induced by LPS [31]. In hemorrhagic shock however, blood withdrawal led to significantly increased liver (but not kidney) and plasma levels of H₂S, and pretreatment of rats with PAG or β-cyanoalanine (BCA; a reversible inhibitor of CSE) reduced liver H₂S synthesis and partially restored systemic blood pressure [20,35].

A role for H₂S in the acute inflammatory aspects of septic and hemorrhagic shock is also evident. For example, PAG treatment of LPS- and CLP-treated animals decreased lung and liver myeloperoxidase (MPO) expression and activity, ameliorated histological lung and hepatic damage and lowered plasma levels of TNF-α and nitrite/nitrate (an index of ^•NO synthesis) [21,29–31]. Similarly, pretreatment of rats with PAG for 1 h prior to blood withdrawal was also protective as PAG increased heart rate recovery time, significantly reduced liver, lung and plasma levels of TNF-α and IL-6 and, in the liver, PAG reduced neutrophilic accumulation and tissue damage [35]. Conversely, administration of H₂S donors based on sulfide salts (see later), such as sodium hydrosulfide (NaSH) to animals either alone or in combination with LPS, induce marked hypotension, liver, lung and kidney inflammation and aggravated multiple organ injury through upregulation of tissue NF-κB, p38 and ERK1/2 signaling [36–38].

In summary, the aforementioned studies suggest that in septic and hemorrhagic shock, induction of endogenous H₂S synthesis is detrimental, since PAG and BCA (CSE inhibitors; see later) reduced inflammation and multiple organ dysfunction in septic shock, and were protective in hemorrhagic shock, and the H₂S-donor NaSH promoted inflammation and tissue injury.

H₂S & inflammatory swelling (edema)

Edema is the excessive accumulation of serous fluid in the intercellular spaces of tissues in skin or in one or more cavities of the body, which occurs as part of the acute inflammatory response because of increased blood vessel wall permeability. Inhalation of H₂S gas is known to induce substantive pulmonary edema, suggesting endogenous H₂S could also promote tissue swelling. Edema is commonly modeled in the laboratory through the use of carrageenan, a mucopolysaccharide extract of Chondrus crispus that induces an acute, defined and nonimmune inflammatory response [39]. Injection of carrageenan into the hindpaw of rats induced a significant increase in H₂S synthesis and neutrophil infiltration, whereas pretreatment with PAG (a CSE inhibitor) resulted in dose-dependent inhibition of hindpaw edema and MPO activity [40]. In sharp contrast to this, a more recent study [41] showed the H₂S ‘donors’ NaSH, sodium sulfide (Na₂S) and Lawesson’s reagent reduced leukocyte infiltration and adherence to the vascular endothelium via glibenclamide-inhibitable K_ATP channels, and the CSE inhibitor BCA promoted leukocyte infiltration. However, the effects of PAG on hindpaw H₂S levels were not determined in this study [41]. A closer examination of the experimental protocol used in both studies suggests inter-laboratory methodological differences could have accounted for these apparent contradictions. For example, Bhatia et al. treated rats with 2% w/v solution of carrageenan for 3 h and added PAG intraplantar into one hindpaw 1 h before carrageenan treatment [40], whereas Zanardo et al. performed time-course studies and used a 1.5% w/v solution, injecting BCA (or H₂S donors) intraperitoneally 30 min prior to carrageenan treatment [41]. It is also possible that BCA and/or PAG exhibited effects independent of CSE inhibition, which modulated hindpaw fluid homeostasis (see later). Therefore, at present, the precise role of H₂S in mediating edema is unresolved.

Role of H₂S in gastrointestinal inflammation

The class of drugs known as NSAIDs, such as aspirin, diclofenac, ibuprofen and celecoxib, are among the most widely prescribed medications in the world and, owing to their antipyretic, analgesic and anti-inflammatory properties, they are used in the treatment of a wide range of conditions, such as headache, migraine, rheumatoid arthritis, osteoarthritis, acute gout, pyrexia and dysmenorrhea. Their primary mode of action is the inhibition of prostaglandin production from the enzymes COX-1 and -2. In the gastric mucosa, prostaglandins (PGs), such as PGE₂, PGI₂, PGF_2a and PGD₂, mediate many of the components of ‘gastric mucosal defense’ [42,43] and maintain gastric blood flow, regulate the secretion of bicarbonate and mucus by surface epithelial cells and repair superficial mucosal injury through epithelial restitution [42,43]. However, while inhibition of COX activity is useful systemically in reducing pain, edema and inflammation, inhibition of COX-1-mediated prostaglandin synthesis in the gastric mucosa impairs blood flow and induces superficial injury. This subsequently triggers an acute inflammatory response, ultimately leading to gastric ulceration (explicitly reviewed in [43]) in 1–4% of patients taking NSAIDs chronically [44].

Recently, modulation of endogenous gastric H₂S synthesis has been suggested as one potential mechanism by which NSAIDs induce gastrotoxicity [45]. CSE and CBS are variably expressed throughout the stomach, jejunum and colon of rats and mice, and low levels of CSE and CBS are present in human colon biopsies, and experiments with CSE and CBS inhibitors showed both enzymes to readily synthesized H₂S [23]. Further immunohistochemical studies showed CBS immunostaining was primarily localized to cells within the muscularis mucosa, submucosa, lamina propria and fibroblasts, whereas CSE staining was more diffuse, possibly indicating an association with blood vessels [23,46]. NSAIDs reduced the expression of CSE and H₂S production in the gastric mucosa, induced histological lesions [41,45] and potentiated acetic acid-induced gastric ulcer formation [47]. Conversely, NaSH induced gastroprotective effects, such as reducing NSAID-induced adherence of leukocytes to the vascular endothelium and reducing mucosal neutrophil infiltration. NaSH also decreased the expression of TNF-α, intercellular adhesion molecule 1 (ICAM-1) and lymphocyte-associated antigen-1, but no significant effect on mucosal PGE₂ synthesis was observed [45]. Interestingly, the K_ATP channel antagonist glibenclamide and the CSE inhibitor PAG (see later) aggravated NSAID-induced gastric injury, whereas pinacidil (a K_ATP channel agonist) reduced gastric injury [45]. Although l-cysteine enhanced ulcer healing, an effect prevented by PAG, this was not inhibited by glibenclamide, suggesting that the presence of protective molecular pathways is independent of K_ATP channel activation in this model [47].

Impaired synthesis of H₂S may also play a role in the development of inflammatory bowel disease and, in particular, ulcerative colitis, an idiopathic chronic inflammatory disorder limited to the colon that manifests as painful colonic ulcers. Recently, in the trinitrobenzenesulfonic acid (TNBS) rat model, Wallace et al., showed that during the development of colitis, levels of colonic H₂S synthesis initially increased in response to TNBS, then decreased in parallel to colonic damage where H₂S synthesis in the colon was mediated by both CSE and CBS enzymes [46]. Furthermore, pharmacological inhibition of CBS and, to a lesser extent, CSE inhibition exacerbated colitis. By contrast, intracolonic administration of NaSH or Lawesson’s reagent (an additional H₂S donor) decreased TNBS-induced colonic tissue damage and thickness and lowered TNF-α expression. Furthermore, treatment of rats with aminooxyacetic acid (AOAA), PAG and the K_ATP channel antagonist glibenclamide substantially increased mortality rates but this was not observed with the BCA (an additional CSE inhibitor) or pinacidil. The disparate effects of PAG and BCA could possibly be due to dosing, the fact that PAG binds to CSE irreversibly [48], whereas BCA binding is reversible [49], or that PAG (and AOAA) may induce effects independent of CSE inhibition (see later). However, collectively, these studies strongly suggest a potential role of H₂S in regulating gastric inflammatory response and microvascular blood flow via K_ATP channel activation.

H₂S & pancreatic inflammation

Both CSE and CBS are highly expressed in pancreatic acinar cells under basal conditions [50]. Upon experimental induction of acute pancreatitis with the decapeptide secretagogue cerulein, acinar cell and lung H₂S synthesis and CSE expression were significantly increased whereas CBS levels were reduced, suggesting CSE as the predominant source of H₂S in this model [50]. Cerulein also elevated plasma H₂S levels. Inhibition of CSE in acinar cells [50] or in cerulean-treated mice [51] with PAG treatment reduced cerulein-induced formation of H₂S in the lung and pancreas and increased expression of the macrophage chemokines MCP-1, MIP-1a and MIP-2, as well as significantly suppressing the proinflammatory neuropeptides substance P (SP), preprotachykinin-A (PPT-A) and neurokinin-1 receptor (NK1R), mediated, at least in part, by the activation of the transcription factor NF-κB [52]. By contrast, the exposure of isolated acinar cells to NaSH (10–100 µmol/l) induced the synthesis of SP, PPT-A and NK1A [50], activated the SFK and caused an upregulation of ICAM-1 [52].

Collectively, the aforementioned studies have strongly suggested that H₂S is proinflammatory in acute pancreatitis. However, the effects of H₂S on the exocrine pancreas are complex. For example, in addition to stimulating the synthesis of proinflammatory mediators, NaSH also induces acinar cell apoptosis (1–100 µmol/l) through JNK/p38 and mitochondrial-dependent pathways [53,54], suggesting that the aforementioned effects on proinflammatory signaling could be due to cell death. Low doses of NaSH were recently shown to reduce cerulein-induced neutrophil accumulation and histological tissue damage in the pancreas and lung, with concomitant downregulation of the pancreatic and pulmonary cell adhesion molecules P-selectin, E-selectin, ICAM-1 and VCAM-1 [55]. The precise reasons for this discrepancy are unknown.

H₂S & skin inflammation (burn injury)

To date, two studies reporting conflicting results have described a potential role for H₂S in acute burn injury using anesthetized mice subjected to either 30 [56,57] or 40% [57] total body surface area (TBSA) full thickness burn injury. TBSA produces a large systemic inflammatory reaction characterized by leukocyte activation and plasma leakage in the microvasculature of tissues and organs remote from the wound. The degree of inflammatory response correlates directly with the percentage of TBSA burn [58,59]. In mice subjected to 30% TBSA burn injury [56,57], significantly elevated plasma and hepatic H₂S levels with a concomitant increase in liver and lung expression of CSE were observed 8 h post-burn injury. In this study, prophylactic and therapeutic administration of PAG reduced burn-associated neutrophil accumulation and histological changes in liver and lung tissues. Injection of NaSH (10 mg/kg intraperitoneally) at the same time as burn injury aggravated burn-associated inflammation to a small, but significant, extent, suggesting H₂S promoted tissue damage and inflammation. By sharp contrast, in animals receiving a 40% TBSA burn injury and smoke inhalation, administration of Na₂S (2 mg/kg subcutaneously) immediately after injury increased survival, decreased lung oxidative stress and substantially reduced lung levels of the proinflammatory cytokine IL-1β but increased lung levels of the anti-inflammatory mediator IL-10 [57]. However, the effects of PAG were not investigated in this study. The precise reasons for these disparate findings are unclear. In one study, Na₂S was administered as a pH- and osmolarity-balanced solution [57], whereas in the other study, NaSH was used but the disparity was unlikely to be due to the donor employed, since once administered to the animals they would give approximately equal total H₂S [55]. However, the route of administration, either intraperitoneal [56] or subcutaneous [57], could reflect significant differences in H₂S bioavailability at the local sites of inflammation, which, in turn, might modulate local inflammatory signaling. A further possibility is an additive effect with subcutaneously injected analgesics. For example, buprenorphine was injected subcutaneously in the study by Esechie et al.[57], while no analgesic was adminstered in the study by Zhang et al.[56]. In summary, the role for H₂S in burn injury is unclear.

H₂S & inflammatory lung disease

Acute respiratory distress syndrome (ARDS) and the less-severe acute lung injury (ALI) are serious reactions to various forms of injuries to the lung (e.g., alcohol or drug abuse, trauma and burns) or acute illness (e.g., sepsis, acute pancreatitis and pneumonia). ARDS is characterized by inflammation of the lung parenchyma leading to impaired gas exchange, release of inflammatory mediators, inflammation, hypoxia, shortness of breath, tachypnea and often multiple organ failure [2]. In the oleic acid (OA) model of ALI, OA decreased arterial PaO₂, increased ALI disease score (index of quantitative assessment [IQA]), polymorphonuclear cell infiltration in the lungs and increased plasma and lung levels of H₂S, IL-6, IL-8 and IL-10. By sharp contrast, injection of NaSH (56 µmol/l) 30 min prior to OA significantly reduced the extent of lung injury, lowered lung and plasma IL-8 and IL-6 levels and reduced inflammatory cell accumulation within 2 h. NaSH also increased lung and plasma IL-10 levels. Although no clear mechanisms were addressed and effects of CSE and CBS inhibitors were not investigated, it is clear that lower plasma (or lung) H₂S levels were indicative of lung injury in this model, and strategies to raise lung H₂S levels may be beneficial in ALI and ARDS.

In addition to a potential role in acute inflammatory conditions as summarized above, a role for H₂S in chronic inflammation is emerging. H₂S gas inhalation is well known to induce substantive pulmonary edema characterized by extensive eosinophilic extravasation into the bronchoalveolar space and fibrinocellular alveolitis [59,60], suggesting modulation of endogenous H₂S synthesis could represent a novel pathway in chronic inflammatory lung disease. Chronic obstructive pulmonary disease (COPD) is characterized by chronic inflammation throughout the airways, parenchymal and pulmonary vasculature, resulting in substantial airflow obstruction. Using a sulfide-specific electrode, Chen et al. showed significantly higher serum H₂S levels in patients with COPD compared with age-matched control subjects or patients with acute exacerbation of COPD [61,62]. Levels of H₂S decreased with progressively worsening respiratory failure. In these studies, serum H₂S levels negatively correlated with sputum neutrophil counts but positively correlated with sputum lymphocyte [60] and macrophage [62] counts, lung function (forced expiratory volume in 1 s) [61,62] and serum ^•NO, suggesting that during COPD, H₂S synthesis is induced in the lung to promote bronchial smooth muscle dilatation to improve airflow and limit inflammatory signaling.

In asthma, chronic inflammation of the airways leads to characteristic airway remodeling including subepithelial fibrosis and airway smooth muscle hypertrophy, resulting in irreversible narrowing of the airways (bronchoconstriction) [63]. A role for H₂S in asthma has also been investigated [64]. In normal rat lung tissue, CSE expression is located primarily in airway and vascular smooth muscle cells [64]. In the ovalbumin (OVA) challenge model of asthma in mice, OVA treatment reduced CSE expression and decreased pulmonary tissue H₂S synthesis. Exogenous administration of NaSH reduced airway inflammation and decreased eosinophilic and neutrophilic influx. NaSH also attenuated OVA-induced pulmonary iNOS activation, reduced airway remodeling (goblet cell and smooth muscle cell hyperplasia and collagen deposition score) and restored peak expiratory flow (PEF). Furthermore, H₂S concentrations in serum and lung positively correlated with PEF and negatively correlated with bronchoalveolar neutrophil and eosinophil count. These studies strongly suggest that induction of CSE-derived H₂S during asthma plays an anti-inflammatory and antiremodeling role in the pathogenesis of asthma.

Although the precise molecular mechanisms by which H₂S induces vascular remodeling and modulates inflammatory signaling in the lung are unknown, therapeutic modulation of endogenous H₂S synthesis may be a viable approach for the treatment of chronic inflammatory lung diseases such as asthma and COPD.

H₂S & inflammatory joint disease

Rheumatoid arthritis (RA) is the most common form of chronic inflammatory joint disease, affecting approximately 1% of the adult population in the Northern hemisphere [65]. As such, RA is of considerable economic concern. RA is characterized by diffuse cartilage loss and the erosion of bone and cartilage. This process is believed to start in the synovial membrane with initial edema and hyperplasia of the synovial lining. Overproliferation of synoviocytes and macrophages causes thickening of the synovial lining, which together with lymphocytes, plasma cells and mast cells, develops into a sheet of invasive cellular tissue that is continuous with the synovial lining (pannus) [4]. As a result of the higher proportion of synoviocytes, the pannus causes erosion of bone and cartilage at the margin of joints, resulting in the loss of cartilage-producing articular chondrocytes, joint erosion and eventual loss of joint mobility [65,66].

Recently, we showed that synovial fluid (SF) aspirates from RA patients contained up to fourfold-higher concentrations of H₂S than in paired plasma samples (RA SF median concentration: 62.41 µmol/l) and more than twofold-higher H₂S levels than SF aspirates from patients with noninflammatory arthritides (median: 25.1 µmol/l) [22]. Furthermore, SF H₂S concentrations negatively correlated with SF neutrophil and total white cell counts, and positively correlated with tender joint counts, suggesting SF H₂S could represent a novel index of disease activity in the inflamed joint. Increased synovial synthesis of H₂S may not be specific to RA since other inflammatory joint diseases such as psoriatic and reactive arthritides also have significantly higher SF levels of H₂S compared with matched plasma (Figure 1). However, whether the increase in synovial H₂S synthesis represents an endogenous mechanism to promote inflammation or whether synthesis was elevated in response to joint injury as an anti-inflammatory response is unknown. Recently, Kloesch et al. showed that in synviocytes isolated from RA patients, administration of a bolus of NaSH transiently decreased and then increased IL-1β-induced synthesis of IL-6 [67]. Mechanistic studies suggested that this was independent of NF-κB activation but dependent on ERK1/2 deactivation. However, it should be noted that the predominant effects were seen at NaSH concentrations equal to or greater than 125 µmol/l (often 1 mmol/l), which are considerably higher than the reported levels of H₂S in synovial fluid [22]. Nevertheless, the same biphasic effect of H₂S on inflammatory signaling has also been observed in LPS-treated murine macrophages where low concentrations of H₂S inhibited LPS-induced synthesis of PGE₂, ^•NO, IL-1β and IL-6 and NF-κB activity, but higher concentrations of NaSH promoted the synthesis of proinflammatory mediators [68]. Furthermore, in an in vivo murine model of acute arthritis induced by kaolin/carrageenan [69], a bolus addition of 30–50 µM of Na₂S inhibited leukocyte adhesion in postcapillary venules in acutely inflamed mouse knees. In contrast to the effects of H₂S on the systemic vasculature, Na₂S induced vessel constriction rather than vasodilatation. Although the precise reasons for this observation are unclear, one possibility is that H₂S induces vessel constriction to counteract local proinflammatory vasodilatory mediators such as PGE₂, ^•NO and histamine. Together, these in vivo and in vitro studies would suggest that increased H₂S synthesis in the inflamed human joint [22] could represent a novel endogenous mechanism to control, limit or resolve inflammation, although further work is clearly required to substantiate this.

Role for H₂S in neurogenic inflammation?

During neurogenic inflammation, the direct stimulation of afferent sensory nerves induces the release of bioactive inflammatory mediators, such as SP, NK-A and calcitonin gene-related peptide (CGRP). SP binding to NKR1 induces local vasodilatation, increased microvascular permeability and edema resulting in the accumulation of leukocytes. H₂S has also been proposed as a novel mediator of neurogenic inflammation. Intraperitoneal injection of NaSH into ‘normal’ mice induced significant NKR1-dependent, but NKR2- and CGRP-independent, increases in plasma levels of SP and pronounced lung inflammation [70], and in sepsis-induced lung injury [71] and acute pancreatitis [72], H₂S upregulated plasma levels of SP and aggravated lung inflammation. These effects were blocked by either a selective NK1R antagonist or genetic depletion of PPT-A. In isolated murine pancreatic acinar cells, NaSH induced, but PAG inhibited, SP release and NK1R/PPT-A gene expression. Furthermore, in the isolated guinea pig airway, NaSH induced the release of SP and CGRP from sensory nerve terminals, and led to airway constriction through capsaicin and NKR1/NKR2-transient receptor potential vanilloid 1 (TRPV1)-dependent pathways [73,74]. Similarly, in isolated rat bladder [75] and rat detrusor muscle [76], NaSH induced tachykin release from caspaicin-sensitive primary afferent neurons, leading to smooth muscle contraction. Collectively, these studies suggest that induction of neurogenic inflammation could mediate the reported proinflammatory role of H₂S in inflammatory pathologies.

Antioxidant activity of H₂S?

It has been long established that increased production of reactive oxygen species and at sites of acute and chronic inflammation is an important component of the inflammatory process [4]. The resultant oxidative stress is an important component of the inflammatory process. For example, upregulation of ^•NO from iNOS, superoxide (O₂^•-) from endothelial cell and phagocytic cell NAD(P)H oxidase (respiratory burst), peroxynitrite (ONOO^-) formed from the interaction of ^•NO with O₂^•-, hydrogen peroxide (H₂O₂) and hypochlorous acid (HOCl), derived from neutrophil MPO, are targeted towards the initial inflammatory insult, for example, bacterial infection. Overproduction of these reactive intermediates would not only damage the bacteria but would damage surrounding tissue, precipitating a further inflammatory response. Inhibition of ‘oxidative stress’ by inactivating enzymes or the direct ‘scavenging’ of the oxidant species would be expected to contribute to the many potential anti-inflammatory effects of H₂S. In vitro experiments have shown that H₂S generated from sulfide salts (Na₂S and NaSH) inhibits cellular damage and intracellular protein oxidation induced by HOCl [77,78], ONOO^-[79] and ^•NO [34,80]. NaSH is also reported to scavenge and/or degrade lipid peroxides [81,82] and inhibit the expression and activity of NAD(P)H oxidase [83,84]. Increased hepatic glutathione (GSH) synthesis and decreased lipid peroxidation are also observed with Na₂S treatment in a murine hepatic ischemia–reperfusion injury model [85]. In neuronal cells, NaSH inhibited cell death induced by β-amyloid [86] and MPP⁺[87], mediated, at least in part, via antioxidant effects, and upregulated intracellular GSH synthesis through increasing cysteine uptake and elevation of glutathione-S-transferase activity [88,89]. Furthermore, administration of Na₂S to mice or overexpression of CSE increased Nrf-2 signaling [90,91]. There is also emerging evidence for antioxidant activity in vivo in acute burn injury [57] and myocardial injury [90,91].

Are there ‘selective’ inhibitors of H₂S synthesis?

A considerable amount of data on the potential role of endogenous H₂S in inflammatory signaling has been obtained by the use of pharmacological inhibitors of CSE (e.g., DL-PAG and BCA) and inhibitors of CBS (e.g., AOAA) (Figure 2). While undoubtedly initially very useful pharmacological tools in a new but rapidly expanding research field, it is unlikely that these compounds are entirely specific since they target the pyridoxal phosphate (PLP) binding site of CSE and CBS [92] and other PLP-dependent and -independent enzymes (summarized in Table 1). This problem is exaggerated by the disparate concentration ranges commonly used in the laboratory (often used up to 10 mmol/l [93–95]). Detailed pharmacokinetic analyses of these compounds have not been determined and the effective concentrations of these inhibitors are unknown. 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, over this concentration range, PAG and AOAA will exert significant effects independent of CSE or CBS inhibition. For example, in the rat colon, BCA induces submucosal edema and hypertrophy of the muscularis [46], whereas PAG inhibits cardiac L-alanine transaminase activity [96–98], and chronic dosing in rats led to increased heart, liver and spleen sizes [99]. Similarly, AOAA has been widely used as a general inhibitor of transamination reactions [97,98], inhibits the malate–aspartate shuttle, preventing lactate gluconeogenesis [100,101], inhibits mitochondrial and cytosolic aspartate transaminase activity [102,103] and inhibits protein synthesis [104]. AOAA may exert at least some of its inhibitory effects on H₂S metabolism through inhibition of 3-mercaptopyruvate sulfur transferase (3-MST) [102,105], an additional enzymatic neuronal [106] and vascular [107] source of H₂S. Although PAG rapidly accumulates in blood and through the D-amino acid oxidase pathway, it is metabolized to an as yet unidentified metabolite, leading to renal injury and significant proteinuria, glucosuria and polyuria [108]. However, this was not observed with L-PAG as opposed to D-PAG. Since the kidney is the primary organ for fluid homeostasis, the renal and diuretic effects of this PAG metabolite (or PAG itself) are likely to contribute to (or confound) any experimental observations, in particular when assessing the precise role of CSE-derived H₂S in swelling and tissue edema. These CSE-independent effects may offer at least one explanation for the disparate effects of H₂S inhibitors and H₂S donors in inflammation. Although AOAA, BCA and PAG clearly inhibit H₂S synthesis in isolated tissue extracts, the true extent to which these compounds are able to penetrate cellular membranes and inhibit intracellular CSE and CBS is yet to be determined. It is possible that there is inconsistent cell penetration and cell metabolism among cell types and animal species, further contributing to the disparity in the literature. However, currently, there are no wholly specific CSE or CBS inhibitors available to researchers. Therefore, experimental data should be interpreted with the aforementioned observations in mind.

H₂S donor molecules & inflammation: notes of caution

As with any new and emerging field of study, researchers are limited by the availability of specific, purpose-designed experimental tools. The vast majority of studies on H₂S, in any area of physiology, as discussed previously, have utilized Na₂S and NaSH as sources of H₂S (Figure 3). While these salts are undoubtedly convenient and useful in that solutions of H₂S (and HS^-) can be readily prepared in the laboratory without the requirement for the use of awkward H₂S gas cylinders, they are not particularly relevant tools to examine the physiology of H₂S in vitro or in vivo. The addition of Na₂S, NaSH or saturated solutions of H₂S gas to aqueous solutions results in the instantaneous release of a bolus of H₂S (and HS^-), which dissipates in seconds [33,68]. It is highly unlikely that cells or tissues are ever exposed to H₂S generated in such a rapid manner, generating high local concentrations of H₂S, since endogenously produced H₂S through CSE and CBS is relatively slow and sustained [68,92,109,110]. Furthermore, injection of these salts into animals results in a minimal increase in plasma H₂S concentrations over a very short period of time [21,33,68,111]. Commercially available NaSH rarely exceeds 70% purity and it is possible that the disparate physiological findings with NaSH could reflect the varying amounts of chemical impurities present in the NaSH. Commercial Na₂S, on the other hand, is considerably purer but in in vitro experiments, still generates H₂S too quickly to be physiologically equivalent. Nonetheless, as a therapeutic tool, pharmaceutical-grade Na₂S shows great potential and has been successfully used to prevent and treat myocardial infarction, ischemia–reperfusion injury and endotoxic shock [112,113]. It is unclear whether these effects are directly due to released H₂S or whether H₂S is sequestered into biologically active intermediaries (e.g., of plasma protein or lipids; also reviewed in detail elsewhere in this issue), further highlighting the complex nature of H₂S physiology and medicine.

Future developments: slow-release H₂S donor molecules

The problem of using rapidly generated H₂S from a bolus of concentrated sulfide salt is currently being addressed. Recently H₂S donors (Figure 3) have been synthesized to circumvent this problem using 5-(4-hydroxyphenyl)-3H-1,2-dithiole-3-thione (ADT-OH), derivatives of existing NSAID molecules, such as indomethacin (ATB-343), diclofenac (ACS-15), naproxen (ATB-345) and mesalamine (ATB-429), reviewed in detail elsewhere in this issue. These derivatives (Figure 3) have shown great promise in alleviating and limiting gastrointestinal side effects and toxicity of NSAIDs and in the treatment of experimental inflammatory bowel disease and arthritis (summarized in Table 2).

However, it should be noted that ADT-OH is the major metabolite of 3H-1,2-dithiole-3-thione (ADT; a dithiolethione), a putative chemopreventative agent structurally related to olitipraz. Dithiolethiones and related compounds have previously been explored as inhibitors of carcinogenesis and have shown considerable promise in animal models of edema and gastrointestinal inflammation (extensively reviewed in [114,115]; summarized in Table 2). However, it is possible that some of the observed biological effects associated with ADT-OH derivatives were due to the dithiolethione moiety itself, rather than released H₂S. Indeed, an examination of the pertinent literature shows that dithiolethiones, including ADT-OH, are reactive with thiols and elicit several effects that may not be due to H₂S. For example, Isenberg et al.[116] and Moody et al.[117] recently showed that the dithiolethione moiety of ADT-OH-modified NSAIDs and valproate rather than released H₂S inhibited angiogenesis and cell proliferation. Others have shown dithiolethiones to induce the synthesis of phase II enzymes (e.g., glutathione-S-transferase, NAD[P]H, quinone oxidoreductase), GSH reductase and catalase [118–122]. Similar to ADT, ADT-OH itself is also expected to elicit antioxidant activity in vivo. The dithiolethione derivatives are convenient, promising in animal studies of pathology as novel therapeutic agents, and represent a considerable advance in studying the physiology, pharmacology and pathology of H₂S. However, care should therefore be taken to interpret experimental data with the above observations in mind. Control experiments consisting of the dithiolethione alone, parent compound, parent compound with ADT-OH and ‘decomposed’ donor molecule should be thoroughly evaluated and the release of H₂S under each experimental condition should be measured.

More recently, a H₂S donor that does not consist of a structurally modified NSAID molecule, GYY4137 [32,33,68], has been synthesized and characterized (summarized in Table 2). GYY4137 is a very slow-releasing H₂S donor compound that releases two molecules of H₂S per molecule of GYY4137 and has been shown to exert prominent endothelium-dependent vasodilatory activity in vivo via K_ATP channel-dependent mechanisms in hypertensive rats [33]. GYY4137 also exerts prominent anti-inflammatory activity in endotoxic shock [32] and in isolated inflammatory cells [68], mediated in part via inhibition of NF-κB/AP-1-dependent proinflammatory signaling (summarized in Table 2). In contrast to ADT-OH derivatives, GYY4137 is freely soluble in water and, unlike S-diclofenac [123], GYY4137 is apparently not toxic to vascular cells or anesthetized rats, even when tested at mmol/l concentrations [32,33]. GYY4137 offers an additional advantage to the researcher over ADT-OH compounds in that its decomposition products appear inactive [68] allowing one to study the physiological effects of slow release of H₂S in inflammation directly, in the absence of any possible effects of NSAID or ADT-OH. The precise metabolic profile for GYY4137 has yet to be elucidated so it is possible that at least some of the reported effects in vivo could be due to metabolism to biologically active intermediates.

Expert commentary & five-year view: is H₂S pro- or anti-inflammatory?

Enzyme inhibitors, such as PAG, AOAA and BCA and sulfide salt H₂S donors (e.g., Na₂S and NaSH), have been useful in highlighting the potential role of a previously unrecognized endogenous biologically active gas as a novel mediator of inflammatory signaling. The bulk of the studies that have used either Na₂S or NaSH as the source of H₂S, or have used PAG (and BCA) and AOAA to inhibit CSE and CBS activity, have generally come to the conclusion that endogenous H₂S is proinflammatory (summarized in Table 3). However, Na₂S and NaSH generate H₂S too quickly to be an effective model for endogenously synthesized H₂S. While PAG, BCA and AOAA decrease blood and tissue levels of H₂S, they are not wholly specific, detailed pharmacodynamic and pharmacokinetic profiles have yet to be determined and the ‘effective’ concentration of drug in plasma and/or tissues is not known. Potential effects of these inhibitors, independent of H₂S synthesis, especially in vivo, also cannot be ruled out. In order for the field to progress, pharmacological tools to selectively inactivate or inhibit CSE and CBS are required. Genetic approaches to remove CSE have made significant progress in unraveling the role of H₂S in regulating blood pressure [13] – although this is also an area of controversy [124] – and these genetic tools have yet to be applied to inflammatory models. Similarly, overexpression of CSE and/or CBS in animals will also shed light on the role of endogenous H₂S in regulating, driving or inhibiting inflammation.

Newer approaches to study H₂S using ADT-OH derivatives or GYY4137 have thus far exclusively shown in a number of cellular and in vivo animal models (Table 2) that slow release of H₂S, which more closely models H₂S derived from CSE or CBS, is anti-inflammatory. This suggests that the observed effect of H₂S in an experimental system may be dependent upon the manner in which cells and tissues are exposed to H₂S. However, even slow-release H₂S donors represent pharmacological H₂S. By analogy to the ^•NO research field, it appears that we are still in the early stages of understanding the true significance of H₂S in the inflammatory response.

In conclusion, the precise role of H₂S in the inflammatory process is ambiguous (summarized in Table 4). There is a great deal of disparity in the literature concerning the anti-inflammatory effects (‘the good’), and the proinflammatory effects (‘the bad’) of endogenous and pharmacological H₂S. As with any new and emerging field, researchers are limited by the availability of suitable experimental tools. In particular, progress has been hampered by a lack of availability of enzyme-specific inhibitors and physiologically irrelevant H₂S donors, which have often been used at very high concentrations (‘the ugly’). The field is further impeded by the current lack of detailed pharmacodynamic and pharmocokinetic profiling of donors and inhibitors. In sharp contrast to the use of sulfide salts (e.g., Na₂S and NaSH) as a source of pharmacological H₂S, the development of novel donors that can deliver H₂S in a slow and sustained manner to model enzymatically synthesized H₂S has universally shown anti-inflammatory activity (‘the promising’). These new molecules offer the potential to clarify the precise role of H₂S in acute and chronic inflammation, as well as to provide an additional class of therapeutic compounds for other emerging aspects of H₂S physiology.

Table 1. Examples of the reported effects of commonly used inhibitors of H₂S synthesis in vitro and in vivo, highlighting the nonspecific nature of these compounds.

Compound	Target	Additional effects/comments	Ref.
PAG	Commonly used as an irreversible inhibitor of CSE	Inhibition of L-alanine transaminase, a commonly used diagnostic enzyme for liver function and a critical enzyme in the hepatic alanine–glucose cycle	[96]
		D-PAG is metabolized to a toxic renal metabolite by a D-amino acid oxidase pathway leading to significant proteinuria, glucosuria and polyurea in rats. This was not observed with L-PAG. The vast majority of in vivo studies examining the role of CSE in inflammation (and other aspects of physiology and pathology) have generally used racemic PAG (DL-PAG) over a wide range of doses (typically 10–100 mg/kg) administered through a variety of routes. Dysregulation of fluid homeostasis induced by renal toxicity will potentially confound studies examining the role of CSE (and H2S) on tissue swelling and edema	[108]
		Induced hepatosplenomegaly in rats, suggesting either PAG or its hepatic metabolism induces a proinflammatory response in the liver and proximal and/or distal tissues	[108]
		PAG decreased plasma levels of taurine and glycine and increased asparagine and citrulline levels indicating that PAG exerts effects on amino acid metabolism independent of CSE, cystathionine, homocysteine and cystine	[136,137]
Aminooxyacetate (O-carboxymethylhydroxylamine hemichloride [CHH])	Commonly used as a reversible inhibitor of CBS	Prevention of lactacte gluconeogenesis by through mitochondrial malate–aspartate shuttle inhibition	[100,138]
		Inhibition of mitochondrial and cytosolic aspartate transaminase	[103,139]
		Inhibition of protein synthesis	[104]
		KATP-dependent and -independent inhibition of insulin secretion in pancreatic β-cells	[140–142]

PAG and aminooxyacetate are widely used as inhibitors of CSE and CBS in vitro and in vivo, respectively. A wide range of concentrations have been employed, typically between 1 and 10 mmol/l. The effective concentrations achieved in vivo are not known and detailed pharmacodynamic and pharmacokinetic analyses have not been determined. Despite this, doses between 30 and 100 mg/kg are generally used either prophylactically or therapeutically. Although these compounds clearly inhibit endogenous H₂S synthesis in vivo, the nonspecific effects highlight the requirement for the development of specific CSE and CBS inhibitors.

CBS: Cystathionine β-synthase; CSE: Cystathionine γ-lyase; H₂S: Hydrogen sulfide; PAG: Propargylglycine.

Table 2. Evidence consistent with an anti-inflammatory role of endogenous and exogenous (pharmacological) H₂S.

Species and model	Compound used	Observation	Mechanism(s)/comment	Ref.
Male Sprague–Dawley rats: Asthma	NaSH	Lower H2S levels in the lung reduces inflammation and airway remodeling	H2S levels in serum and lung correlated with in vivo inflammatory markers (i.e., positive correlation observed with peak expiratory flow) Negative correlation observed with intra-abdominal pressure, eosinophil and neutrophil count, inflammatory cell infiltration, collagen deposition, goblet cell and airway smooth muscle hyperplasia Endogenous generation of H2S did not significantly alter iNOS expression but iNOS negatively correlated with H2S levels in serum and lung tissue, suggesting H2S-mediated inhibition of iNOS NaSH (14 µmol/kg, intraperitoneal) significantly improved lung function (peak expiratory flow, peak inspiratory flow, intrapressure and maximum rising slope of intrapressure) and inhibited iNOS activation in the lung	[64]
Male C57B1/6 mice and male Wistar rats: Acute inflammation of the knee induced by kaolin–carrageenan (monoarthritis model)	Na2S	Intra-articular injection of the H2S donor Na2S inhibited leukocyte- mediated inflammation of the knee	Na2S (50 µmol/l) increased leukocyte velocity, decreased leukocyte rolling flux and decreased leukocyte adherance in post-capillary venules of inflamed mouse knees; inhibited by the KATP channel antagonist glibenclamide Na2S decreased synovial blood flow but had no effect on joint pain	[69]
Female C57BL/6 mice: Acute lung injury; B + S injury	Clinical grade Na2S	Na2S protective against acute lung injury	Na2S (2 mg/kg, subcutaneous) decreased lung levels of the proinflammatory cytokine IL-1β but increased levels of the anti-inflammatory cytokine IL-10 after B + S injury and decreased mortality Na2S decreased protein oxidation and preserved lung ultrastructure	[57]
Male Sprague–Dawley rats: LPS-induced endotoxic shock LPS-treated murine macrophages	GYY4137	Slow-release H2S donor is anti-inflammatory	GYY4137 (10–1000 µmol/l) decreased LPS-induced TNF-α production from rat blood and in anesthetized rats and isolated murine macrophages decreased LPS-induced activation of NF-κB, expression of iNOS, COX-2 and levels of PGE2 and ^•NO GYY4137 (50 mg/kg, intraperitoneal) also inhibited LPS-induced lung and liver injury, decreased lung and liver neutrophil accumulation and elevated plasma levels of the anti-inflammatory chemokine IL-10 These effects were not observed with decomposed (‘spent’) GYY4137 clearly showing the effects of released H2S GYY4137 also decreased LPS-induced hypotension	[32]
In vitro cellular study using murine RAW264.7 macrophages	GYY4137 NaSH	Slow-release of H2S was anti-inflammatory whereas ‘fast release’ of H2S from NaSH was pro-inflammatory	GYY4137 (10–1000 µmol/l) inhibited LPS-induced formation of TNF-α, IL-1β, IL-6, PGE2 and ^•NO but increased the synthesis of the anti-inflammatory chemokine IL-10 through suppression of NF-κB/AP-1 activity NaSH (10–1000 µmol/l) aggravated LPS-induced formation of proinflammatory mediators This study highlights the complexities of slow release of H2S (from GYY4137) with fast release of H2S (from NaSH) on inflammatory signaling	[68]
Balb/C Mice: Acute pancreatitis and associated lung inflammation induced by cerulein	H2S releasing Diclofenac (ACS15)	Slow-release H2S donor was anti-inflammatory	Therapeutic or prophylactic administration of ACS15 (50 mg/kg, intraperitoneal) significantly decreased reduced lung inflammation to a significantly greater extent than diclofenac alone; however, ACS15 was ineffective against pancreatic injury (i.e., acinar cell death, pancreatic neutrophil accumulation or plasma amylase)	[129]
Rats: OA-induced acute lung injury	NaSH	Downregulation of endogenous H2S involved in the pathogenesis of OA-induced acute lung injury	OA-treated rats had higher plasma and lung levels of proinflammatory cytokines (IL-6 and IL-8) and lower levels of H2S and anti-inflammatory cytokines (IL-10) compared with controls Administration of NaSH (56 µmol/l, intraperitoneal) to OA-treated rats decreased IL-6 and IL-8 but increased IL-10 levels	[130]
Male Sprague–Dawley rats: LPS-induced endotoxic shock	ACS15 and NaSH	Slow-release of H2S from ACS15 is anti-inflammatory	ACS15 (47.2 µmol/kg, intraperitoneal) decreased plasma and tissue levels of pro-inflammatory mediators (^•NO, PGE2, TNF-α and IL-1β) and increased anti-inflammatory mediators (IL-10) NaSH (47.2 µmol/kg, intraperitoneal) increased tissue MPO activity and inflammation ACS15 decreased mRNA levels of CSE, CBS, COX-1, COX-2, HO-1 and iNOS ACS15 inhibited LPS-induced activity of proinflammatory transcription factors NF-κB and AP-1 (c-fos)	[131]
Male Wistar rats: Carrageenan-induced paw edema Air-pouch model of localized inflammation	NaSH (Na2S, Lawesson’s reagent, BCA)	H2S is proinflammatory by promoting leukocyte–endothelial cell attachment and activation and suppresses edema via KATP-sensitive processes	Intragastric NaSH (100 µmol/kg), Lawesson’s reagent (0.1–3 µmol/kg), Na2S (1–100 µmol/kg) inhibited aspirin-induced leukocyte adherence to mesenteric venules, possibly via KATPchannel-dependent mechanisms (ED50 of 42.7, 1.3 and 29.9 µmol/kg, respectively), whereas inhibition of endogenous H2S synthesis (via CSE) using BCA (50 mg/kg, intraperitoneal) induced leukocyte adherence Inhibition of CSE with BCA (50 mg/kg, intraperitoneal) prevented leukocyte rolling velocity and adhreance NaSH and Na2S reduced leukocyte infiltration induced in the air pouch by carrageenan (ED50 of 35 and 28 µmol/kg, respectively)	[41]
In vitro cellular study N-formyl-methionyl-leucyl-phenylalanine-stimulated polymorphonuclear cells	Na2S	H2S (albeit at high concentrations) inhibited chemotaxis and degranulation of leukocytes	High concentrations of Na2S (up to 2 mM) inhibited ionomycin-induced intracellular Ca^2+ mobilization and MPO/lactoferrin release in polymorphonuclear cells (see later) Na2S primed polymorphonuclear cells for stronger Ca^2+-dependent and Ca^2+-independent oxidative burst	[132]
In vitro cellular study formyl-methionyl-leucyl-phenylalanine- and zymosan-activated serum stimulated polymorphonuclear cells	Na2S	H2S modestly inhibited (albeit at high concentrations) chemotaxis and degranulation of leukocytes	Modest inhibition of polymorphonuclear cell migration by high (2 mM) concentrations of Na2S No effect of Na2S on MPO or lactoferrin release from polymorphonuclear cells	[133]
Female Balb/c mice: Trinitrobenzene sulfonic acid-induced colitis	S-mesalamine (ATB-429)	S-mesalamine dose-dependently reduced disease acitivity and colonic inflammation	S-mesalamine (33–130 mg/kg, oral) reduced the activity of colonic myeloperoxidase and significantly decreased colonic levels of IFN-γ, TNF-α, IL-1, IL-2, IL-12 and RANTES. By sharp contrast, no effect was observed with mesalamine or ADT-OH treatment alone, suggesting protective effects in addition to H2S release	[111]
Male Wistar rats: Carageenan-induced hindpaw edema	S-diclofenac (ACS-15)	S-diclofenac induced time- and dose-dependent inhibition of carageenan-induced edema	S-diclofenac reduced hindpaw edema, neutrophil accumulation and synthesis of PGE2 and ^•NO more effectively than diclofenac alone (ED30: 14.2 µmol/kg; intraperitoneal) The study concluded that the additional anti-inflammatory effect of S-diclofenac was due to H2S release	[134]
Male Wistar rats: Trinitrobenzene sulfonic acid-induced colitis	PAG BCA AOAA	Inhibition of H2S synthesis exacerbated colitis	CSE and CBS contribute to colonic H2S synthesis during experimentally induced colitis Inhibition of endogenous H2S synthesis with either PAG (50 mg/kg, intraperitoneal), BCA (50 mg/kg, intraperitoneal) or AOAA (30 mg/kg, intraperitoneal) increased colonic inflammation and mortality whereas NaSH or Lawesson’s reagent (each at 30 µmol/kg, intracolonic) reduced colonic damage, colon thickness and TNF-α expression, PAG (3 mmol/l) and AOAA (3 mmol/l) inhibited H2S synthesis in colon biopsies	[135]

AOAA: Aminooxyacetic acid; B + S: Flame burn injury and smoke inhalation; BCA: β-cyanoalanine; CBS: Cystathionine β-synthase; CSE: Cystathionine γ-lyase; HO-1: Heme oxygenase 1; iNOS: Inducible nitric oxide synthase; LPS: Lipopolysaccharide; MPO: Myeloperoxidase; NaSH: Sodium hydrosulfide; OA: Oleic acid; PAG: Propargylglycine; PGE: Prostaglandin E2.

Table 3. Evidence consistent with a proinflammatory role of endogenous and exogenous (pharmacological) H₂S.

Species and model	Compound used	Observation	Mechanism(s)/comment	Ref.
Male Sprague–Dawley rats: Hemorrhagic shock	NaSH PAG BCA	H2S increased plasma cytokines	Pretreatment of animals with PAG (50 mg/kg, intraperitoneal) or BCA (20 mg/kg, intraperitoneal) increased heart rate recovery and decreased hemorrhagic shock-induced increase in plasma TNF-α, IL-6 and iNOS PAG also decreased lung and liver MPO activity	[20,35]
Male Swiss albino mice: Cecal ligation and puncture-induced sepsis	NaSH PAG	H2S promoted neutrophil infiltration	Cecal ligation and puncture increased expression of cell adhesion molecules (ICAM-1, P-selectin and E-selectin) promoting neutrophil infiltration NaSH (10 mg/kg, intraperitoneal) PAG (50 mg/kg, intraperitoneal)	[72]
Male Swiss albino mice: Acute pancreatitis induced by cerulein	NaSH	H2S mediated substance P-induced acute inflammation	Cerulein increased plasma H2S and substance P levels, which were decreased by PAG (100 mg/kg, intraperitoneal) Prophylactic and therapeutic PAG decreased pancreatic H2S synthesis, lung and pancreatic levels of NK1R and PPT-A gene expression NaSH upregulated leukocyte rolling, adhesion, adhesion molecules expression, neutrophil infiltration via MIP-2 and IL-8 receptor expression; attenuated by NF-κB inhibition using BAY 117082. Activation of NF-κB–SP-NK1R pathway proposed to mediate H2S involvement in acute inflammation/pancreatitis	[125]
Male Swiss albino mice: Cecal ligation and puncture-induced sepsis	NaSH PAG	H2S increased proinflammatory cytokine expression	NaSH (10 mg/kg, intraperitoneal) increased cecal ligation and puncture-induced MPO activity, cytokine levels (TNF-α, IL-1β, IL-6, MIP-2 and MIP-1a) and lung microvascular permeability PPT-A gene deletion or pretreatment of animals with the NK1R receptor antagonist L703606 attenuated NaSH-induced lung inflammation Prophylactic or therapeutic treatment with PAG (50 mg/kg, intraperitoneal) decreased PPT-A gene expression and substance P levels whereas NaSH increased substance P levels and lung injury	[71]
Balb/C mice: Acute pancreatitis and associated lung inflammation induced by cerulein	PAG	Blockade of CSE activity with PAG is anti-inflammatory	Therapeutic and prophylactic administration of PAG (100 mg/kg, intraperitoneal) inhibited cerulein-induced pancreatic inflammation and lung injury	[70]
Guinea-pig: Airway inflammation	NaSH	H2S induced neurogenic inflammation in the pulmonary circulation	Infusion of NaSH (10 mM) induced neuropeptide release (substance P, CGRP); attenuated by capsaicin desensitization, TRPV1 antagonism (capsazepine) and NK1 (SR140333) and NK2 (SR48968) antagonism NaSH induced plasma extravazation; inhibited with NK1, NK2 and TRPV1 antagonists NaSH induced bronchial (EC50 1.3 ± 0.13 mM) and tracheal (EC50 0.9 ± 0.2 mM) muscle contraction	[73]
In vitro cellular study Differentiated human U937 monocytes	NaSH	H2S induced pro-inflammatory cytokine expression	Exposure of U937 cells to NaSH (0.1–1 mM) led to rapid degradation of IκB and activation of NF-κB p65, ERK1/2 phosphorylation and increased mRNA and protein levels of TNF-α, IL-1β and IL-6; attenuated by inhibitors of ERK and NF-κB ERK–NF-κB pathway proposed	[126]
Male Sprague–Dawley rats: Inflammatory pain induced by formalin injection	NaSH PAG	H2S is a mediator of inflammatory pain	Formalin injection into hindpaws increased hindpaw but decreased spinal cord H2S levels PAG (50 mg/kg, intraperitoneal) for 1 h prior to formalin inhibited pain-related flinch response and hindpaw edema, whereas NaSH (10 nmol/ml, intraplantar) aggravated formalin-induced flinch response PAG decreased c-Fos expression in laminae I–II and V–VI in spinal cord sections	[127]
Male C57B1/6 mice and male Wistar rats: Mechanical inflammatory hypernociception	NaSH	Dual role in inflammatory nociception propose pronociception due to neutrophil migration and antinociception due to blockade of KATP channels	PAG (10–30 mg/kg, intra-articular, intraplantar [intraperitoneal] and subcutaneous) and NaSH (10–90 µmol/kg, intraplantar) decreased zymosan and LPS-induced mechanical paw hypernociception and decreased neutrophil recruitment; effects prevented by KATP channel antagonist glibenclamide NaSH (90 µmol/kg intraperitoneal) increased MPO activity; PAG decreased MPO activity but no change in plasma cytokines was observed (i.e., TNF-α, IL-1β, IL-8 receptor)	[128]
Balb/C mice: Acute pancreatitis and associated lung inflammation induced by cerulein	NaSH PAG	Pro-inflammatory effect of H2S mediated through chemokines	PAG (50 mg/kg, intraperitoneal) downregulated but NaSH (100 µM) upregulated cerulein-induced increase in MCP-1, MIP-1a and MIP-2 expression	[51]
Male Swiss albino mice: LPS-induced endotoxic shock	PAG NaSH	H2S is pro-inflammatory and mediates LPS-induced tissue in inflammation PAG reduced LPS-induced tissue inflammation and damage	LPS induced time- and concentration-dependent increases in plasma H2S CSE expression and activity increased in liver, kidney and lung Injection of mice with NaSH (14 µmol/kg, intraperitoneal) resulted in marked histological signs of inflammation, increased lung and liver MPO activity and increased plasma TNF-α levels Administation of PAG (50 mg/kg, intraperitoneal) reduced histological lung and liver inflammation and tissue damage as well as reduced neutrophil accumulation	[21]
Male Wistar rats: LPS-induced endotoxic shock	PAG	H2S mediates the pathophysiology of organ injury in endotoxemia	Endotoxemia induced significant increases in CSE and CBS expression, plasma H2S and MPO activity and organ failure (hepatic, pancreatic and neuromuscular injury); effects blocked by prophylactic administration of PAG (50–100 mg/kg, intravenous)	[31]
Male Wistar rats: Carrageenan-induced hindpaw edema	PAG	H2S-induced edema	PAG (25–75 mg/kg, intraperitoneal) significantly and dose-dependently reduced carrageenan-induced hindpaw edema, MPO activity and reduced H2S synthesis	[40]
Male Swiss albino mice: Cecal ligation and puncture model of sepsis	NaSH PAG	H2S upregulated the production of proinflammatory mediators and exacerbated systemic inflammation in sepsis via a NF-κB-mediated process	NaSH (10 mg/kg, intraperitoneal) increased but PAG decreased cecal ligation and puncture-induced NF-κB transcriptional activity PAG (50 mg/kg, intraperitoneal) decreased but NaSH increased cecal ligation and puncture-induced lung and liver mRNA and protein levels of IL-1β, IL-6, TNF-α, MCP-1 and MIP-2; processes antagonized by the NF-κB inhibitor BAY 11–7082	[37]

BCA: β-cyanoalanine; CBS: Cystathionine β-synthase; CSE: Cystathionine γ-lyase; H₂S: Hydrogen sulfide; ICAM: Intercellular adhesion molecule; iNOS: Inducible nitric oxide synthase; LPS: Lipopolysaccharide; MCP: Monocyte chemotactic protein; MIP: Macrophage inflammatory protein; MPO: Myeloperoxidase; NaSH: Sodium hydrosulfide; NK: Neurokinin; NK1R: Neurokinin-1 receptor; PAG: Propargylglycine; TRPV1: Transient receptor potential vanilloid 1.

Table 4. Summary of the effects of exogenous and endogenous H₂S on inflammation and edema.

Inflammatory pathology	Effect of sulfide salts (e.g., fast-release donors, Na2S/NaSH)	Effect of slow-release H2S donors	Effects of H2S- synthesis inhibitors
Sepsis	↑	↓	↓
Edema	↑	↓	↓
Gastrointestinal inflammation: NSAID toxicity Ulceration	↓ ↑	↓ ↓	↑ ↑
Pancreatitis	↑	↓	↓
Burn injury	↓	-	↑
Lung inflammation: Asthma Chronic obstructive pulmonary disease	↓ ↓	- -	↑ ↑
Arthritis	↓	-	-
Neurogenic inflammation	↑	-	↓

↓: Decreases inflammation and edema; ↑: Increases or induces inflammation and edema; -: Not known; H₂S: Hydrogen sulfide; Na₂S: Sodium sulfide; NaSH: Sodium hydrosulfide.

Key issues

• • The pro- and anti-inflammatory actions of hydrogen sulfide (H₂S) involve multiple pathways and are complex.

• • The observed pro- and anti-inflammatory effects of H₂S may be influenced by the exact nature of the model of inflammation (e.g., acute vs chronic inflammation), and the timing of the administration of a H₂S donor or inhibiting compound.

• • The effects of H₂S in models of inflammation are concentration and time dependent, and may involve ‘bell-shaped’ dose–response curves.

• • Slow-release H₂S donors are a more physiologically relevant source of H₂S than sulfide salts.

• • Commonly used inhibitors of cystathionine γ-lyase and cystathionine β-synthase are not wholly specific, and careful control experiments are required to rule-out H₂S-independent effects.

Financial & competing interests disclosure

The authors would like to thank the Wellcome Trust (#091725), the European Union FP7 (RedCat #215009) and the Devon Arthritis Appeal Research Trust for continued financial support. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

No writing assistance was utilized in the production of this manuscript.