Inducible hydrogen sulfide synthesis in chondrocytes and mesenchymal progenitor cells: is H2S a novel cytoprotective mediator in the inflamed joint?

Abstract Hydrogen sulfide (H2S) has recently been proposed as an endogenous mediator of inflammation and is present in human synovial fluid. This study determined whether primary human articular chondrocytes (HACs) and mesenchymal progenitor cells (MPCs) could synthesize H2S in response to pro-inflammatory cytokines relevant to human arthropathies, and to determine the cellular responses to endogenous and pharmacological H2S. HACs and MPCs were exposed to IL-1β, IL-6, TNF-α and lipopolysaccharide (LPS). The expression and enzymatic activity of the H2S synthesizing enzymes cystathionine-β-synthase (CBS) and cystathionine-γ-lyase (CSE) were determined by Western blot and zinc-trap spectrophotometry, respectively. Cellular oxidative stress was induced by H2O2, the peroxynitrite donor SIN-1 and 4-hydroxynonenal (4-HNE). Cell death was assessed by 3-(4,5-dimethyl-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and lactate dehydrogenase (LDH) assays. Mitochondrial membrane potential (DCm) was determined in situ by flow cytometry. Endogenous H2S synthesis was inhibited by siRNA-mediated knockdown of CSE and CBS and pharmacological inhibitors D,L-propargylglycine and aminoxyacetate, respectively. Exogenous H2S was generated using GYY4137. Under basal conditions HACs and MPCs expressed CBS and CSE and synthesized H2S in a CBS-dependent manner, whereas CSE expression and activity was induced by treatment of cells with IL-1β, TNF-α, IL-6 or LPS. Oxidative stress-induced cell death was significantly inhibited by GYY4137 treatment but increased by pharmacological inhibition of H2S synthesis or by CBS/CSE-siRNA treatment. These data suggest CSE is an inducible source of H2S in cultured HACs and MPCs. H2S may represent a novel endogenous mechanism of cytoprotection in the inflamed joint, suggesting a potential opportunity for therapeutic intervention.


Introduction
H2S is a pungent gas recently shown to be endogenously produced in a variety of mammalian tissues primarily from the amino acids cysteine and homocysteine by pyridoxal-5Ј-phosphate-dependent enzymes such as cystathionine-␥-lyase (CSE; E.C. 4.4.1.1) and cystathionine-␤-synthetase (CBS; E.C. 4.2.1.22) (reviewed in [1,2]). Recent studies in animals have identified a role for H2S as a novel gaseous mediator in models of acute and chronic inflammation such as oedema, haemorrhagic and endotoxic shock, asthma, acute lung and burn injury and ischaemia-reperfusion injury (reviewed in [3]). In each of these disease models, tissue CSE expression was up-regulated leading to enhanced H2S biosynthesis and plasma H2S levels. Furthermore administration of sulfide salts such as sodium sulfide (Na2S) or sodium hydrosulfide (NaSH) as crude 'H2S donors' to animals has led to extensive tissue inflammation and oedema as well as increased expression of pro-inflammatory cytokines such as TNF-␣, IL-1␤ and IL-6 through NF-B-mediated processes [4][5][6][7]. These studies have been interpreted as suggesting that H2S is a pro-inflammatory mediator.
Physiological levels of H2S in rodent and human serum and plasma have been reported in the range of 23-45 mol/l (reviewed in [2]). Blood H2S levels are markedly increased in animals by pro-inflammatory mediators such as bacterial endotoxin and carrageenan, and are increased up to fourfold in intensive care patients with sepsis [8]. Recently, synovial fluid (SF) aspirates from patients with rheumatoid arthritis (RA) were shown to contain up to fourfold higher concentrations of H2S than in paired plasma samples (RA SF median concentration, 62.4 mol/l) and more than twofold higher H2S levels than SF aspirates from patients with non-inflammatory arthritides (median, 25.1 mol/l) [9]. However, increased synovial synthesis of H2S may not be specific to RA as other inflammatory joint diseases such as psoriatic and reactive arthritides also have significantly higher SF levels of H2S compared with matched plasma or patients with osteoarthritis (OA) [3].
Despite these observations, the precise role of H2S in inflammatory tissue destruction has not been elucidated. The effects of endogenous and pharmacological H2S on resident joint cells, such as articular chondrocytes, have not been studied in detail and the molecular signalling pathways leading to H2S synthesis in any cell type have not been examined. One recent study using synoviocytes isolated from RA patients [10] examined the effects of pharmacological H2S, generated by sodium hydrosulfide (NaSH). In this study NaSH transiently decreased and then increased IL-1␤-induced synthesis of IL-6 via a mechanism independent of NF-B activation but dependent upon the inactivation of the mean arterial pressure (MAP) extracellular regulated kinase (ERK). However, it should be noted that the predominant effects were seen at NaSH concentrations equal to or greater than 125 mol/l (often 1 mmol/l) which were considerably higher than the reported levels of H2S in SF [9]. Nevertheless, the same biphasic effect of H2S on inflammatory signalling have also been observed in LPS-treated murine macrophages, where low concentrations of H2S inhibited LPSinduced synthesis of PGE2, • NO, IL-1␤ and IL-6 and NF-B activity, but higher concentrations of NaSH promoted the synthesis of pro-inflammatory mediators [11]. Furthermore, in an in vivo murine model of acute arthritis induced by kaolin/carrageenan [12], 30-50 M of Na2S inhibited leucocyte adhesion in postcapillary venules in acutely inflamed mouse knees suggesting an anti-inflammatory role for H2S in this model of arthritis. However, the effects of H2S on cartilage-producing chondrocytes have not been investigated.
The loss of cartilage-producing chondrocytes is a hallmark on both OA and RA pathology. Chondrocytes are a differentiated cartilage-producing cell type derived from MPCs. The pluri-potent nature of MPCs results in the potential generation of several lineages including osteoblasts, adipocytes, myoblasts and tenocytes [13,14]. SF aspirates from OA patients contain immature mesenchymal cells, and normal adult cartilage contains MPCs capable of chondrogenic differentiation [15,16]. Furthermore, increased numbers of these cells are observed in cartilage from OA patients, strongly suggesting a role of MPCs in cartilage repair and pathological cartilage remodelling in various arthropathies. Cartilage-producing chondrocytes are known to undergo apoptotic-like cell death in OA [17,18] and RA ( [19][20][21][22]), an event involving mitochondrial dysfunction [23][24][25], increased oxidative stress [26] and closely correlated to cartilage destruction [18,27,28], pain and inflammation [29]. Because MPCs are also likely to be exposed to the initial cartilage-damaging agent(s) in arthropathies, their fate in the degenerating or chronically inflamed human joint is as important as that of the cells they are replacing. As such, when differentiated into a chondrocytic phenotype, these cells provide a useful model for studying cartilage cell function [30][31][32].
Therefore with these observations in mind, we have investigated whether primary HACs and MPCs, differentiated into a chondrogenic lineage, are able to synthesize H2S in response to pro-inflammatory mediator stimulation. We have investigated the potential physiological and pathophysiological consequences of H2S production by these cells in regulating chondrocyte cell death and determined the effects of a novel slow releasing H2S donor (GYY4137) [33,34] on these cells. For the first time our study shows that endogenous H2S is inducible in chondrocytes and in MPCs, and that endogenous H2S or slowly released 'pharmacological' H2S from GYY4137 is cytoprotective. Our findings also suggest that during joint inflammation H2S may represent a novel endogenous mechanism for preserving joint integrity.

Isolation and characterization of human bone-marrow-derived MPCs
This study was approved by the ethics committee of the National University Hospital and National University of Singapore and informed written consent was obtained from each patient. MPCs were isolated from the trabecular bone chips of patients undergoing reconstructive surgery. After rinsing the sample with saline solution an explant culture system was performed. Colony forming units of spindle-shaped fibroblast-like cells were expanded in two-dimensional culture systems prior to using them for the experiments [30][31][32]. Cells were then differentiated into chondrocytic phenotype as described [30][31][32] in Ham's F-12 media containing 2 mmol/l glutamine, 2 mmol/l dexamethasone, ascorbic acid 2 phosphate (50 g/ml), 1 mmol/l sodium pyruvate, proline (40 g/ml), transforming growth factor ␤-3 (10 ng/ml) and insulin, transferrin and selenium (ITSϩ3; Sigma-Aldrich) at a final concentration of 6.25 g/ml [30][31][32]. Chondrocytic phenotype was confirmed by microscopic evaluation, staining for glycosaminoglycan production (Alcian Blue and Safranin O staining) and collagen type II expression [30][31][32]. We herein refer to these differentiated cells as MPCs. For comparative purposes primary HACs from a donor free from inflammatory or erosive joint disease were obtained from PromoCell (Singapore), expanded in monolayer culture up to passage 5 in chondrocyte growth medium (#C-27101; Promocell), supplemented with Supplement Mix (#C-39635; Promocell) and, unless otherwise stated, embedded in alginate beads as described in [32,35,36]. After treatment of chondrocytes with cytokines or oxidants, calcium alginate was dissolved in 100 mol/l sodium citrate [32,35,36] and cellular viability determined as described later.

Induction of CSE and CBS expression and measurement of cellular H 2 S synthesis
Cells were exposed to the inflammatory mediators TNF-␣; IL-1␤ and IL-6 and added to cell culture media at a final concentration of 5 ng/ml for up to 18 hrs. Initial pilot studies showed that treatment conditions did not induce significant cytotoxicity when assessed by MTT or LDH release assays as described later. Cells were lysed in RIPA buffer (#R0278; Sigma-Aldrich), pH 8.0 and protein expression of CSE, CBS and ␤-actin determined by standard Western blotting as described [37,38]. For antibody detection, an enhanced chemiluminescence detection kit (GE Healthcare, Amersham, Buckinghamshire, UK) was used followed by analysis using a Kodak Image Analyser (IS440CF; NEN Life Science, Boston, MA, USA). Images were captured and analysed using Kodak digital science onedimensional image analysis software. To investigate potential pathways involved in induced CSE/CBS expression and activity, commercially available inhibitors were used and added to cells for 1 hr prior to cytokine or LPS stimulation. Inhibitors of NF-B (PPM-18 and BAY 11-7085; 10 mol/l), p38 (SB203580 and PD169316; 10 mol/l), JNK (SP600125 and AS601245; 10 mol/l) and ERK1/2 (Ste-MEK1(13) and FR180204; 10 mol/l) were used [30].

Assessment of cell death and determination of mitochondrial membrane potential (⌬⌿m)
There is extensive evidence for a role of oxidative and nitrosative stress in inflammatory joint diseases (reviewed in [39]). Pro-inflammatory oxidants such as reactive nitrogen species [26,40] and hydrogen peroxide (H2O2) [41] as well as the products of biomolecule oxidation such as lipid-derived aldehyde 4-HNE [42] are well known to induce apoptotic cell death in human chondrocytes. Similarly, there is increasing evidence for a predominant role of mitochondrial dysfunction in inflammatory and degenerative joint diseases [23,25,43].
Cellular viability was assessed using standard laboratory techniques: trypan blue dye exclusion, reduction of MTT and LDH release assays [44]. MTT data are expressed as percentage of untreated cells, and leakage of LDH into the culture media was measured at 340 nm using a commercially available kit (CytoTox96, Promega) after 18 hrs [26]. Mitochondrial membrane potential (⌬⌿m) in whole cells [45] was assessed using the potentiometric dye tetramethylrhodamine methyl ester (TMRM) and flow cytometry [26,31,38]. Cells were loaded for 30 min. at 37ЊC with TMRM at a final concentration of 50 nmol/l [26,31,38]. Levels of adenosine triphosphate (ATP) were determined with luciferase [44,46] and chemiluminescence measured using an LMax micro-plate reader (Molecular Devices) [44]. ATP levels were normalized to protein content using the Bradford assay (Bio-Rad). Cytosolic and mitochondrial levels of cytochrome c were assessed by ELISA (#APT200; Chemicon, Temecula, CA, USA) according to the manufacturer's instructions and absorbance at 450 nm read using a SpectraMax190 microplate reader (Molecular Devices) [38].

Effects of H 2 S on pro-cell survival signalling
To investigate whether H2S could activate pro-cell survival signalling, cells were treated with GYY4137 or L-cysteine in the presence and absence of CSE and CBS inhibitors. Phosphorylation of Akt (Thr 308 ) and ERK1/2 was assessed by ELISA (Cell Signaling Technology) according to the manufacturer's instructions using 30 g protein. To determine whether activation of PI3K/Akt and ERK was required for protection against oxidative stress-induced cell death, MPC and chondrocytes were treated for 1 hr with inhibitors of Akt {1L-6-Hydroxymethyl-chiro- 6-HO} or triciribine; 5 mol/l), ERK1/2 (Ste-MEK113 or FR180204) or PI3K (LY29402 or wortmannin; 5 mol/l and 10 mol/l, respectively). After this time GYY4137 (100-500 mol/l) was added for a further 1 hr followed by SIN-1, H2O2 and 4-HNE. Cell death was then assessed by MTT and LDH release assays as described earlier.

siRNA-mediated protein knockdown
For siRNA-mediated protein knockdown [31], cells were cultured in monolayers and transfected for 48 hrs in Ham's F-12 media containing CSE siRNA or CBS siRNA (10 nmol/l; Abnova), Akt siRNA, p65NF-B siRNA (80 nmol; Cell Signaling Technology) and a Silencer ® siRNA Transfection Kit (Ambion) according to the manufacturers' instructions. These treatment conditions were identified from preliminary optimization experiments. Non-coding siRNA transfections were also performed as negative controls using two commercially available negative control siRNAs (#4635 and #4611; termed herein as non-coding controls, NCC-1 and NCC-2, respectively; Ambion). Each of these comprised of a 19 base-pair non-targeting sequence with 3Ј dT overhangs and had no significant similarity to any known human gene sequence (Ambion). These transfection conditions were chosen based on preliminary optimization experiments to ensure that siRNA-mediated protein knockdown, siRNA support reagents or transfection conditions did not significantly reduce cell viability (MTT and LDH assays) as described earlier.

Statistical analysis
Data are expressed as mean Ϯ standard deviation of the mean (S.D.) of separate experiments (n Ն 6) performed on separate days using freshly prepared reagents. Where significance testing was performed, ANOVA was used (*P Ͻ 0.05, **P Ͻ 0.01, ***P Ͻ 0.001) and concentration-dependent effects investigated with post-hoc Dunnett's test using SPSS v15.0 software.  Figure 1D-E, respectively. Basal levels of H2S synthesis were significantly inhibited by AOAA (a CBS inhibitor) (Fig. 1F). In contrast, a small statistically insignificant decrease in H2S synthesis was observed after treatment with the CSE inhibitor PAG, suggesting that CBS was the predominant source of H2S under basal conditions. In contrast, treatment of MPCs with the pro-inflammatory cytokines TNF-␣, IL-1␤ and IL-6 significantly increased expression ( Fig. 1A-E) and activity (Fig. 1F) of CSE but not CBS. Incubation of chondrocytes in alginate culture with TNF-␣, IL-1␤ and IL-6 under the same experimental conditions as MPCs also resulted in significant increases in expression ( Fig. 2A and B) and activity (Fig. 2C) of CSE but not CBS. In MPCs and chondrocytes PAG, but not AOAA, significantly inhibited cytokine-induced CSE activity, suggesting that CSE is an inducible source of H2S in these cells.

Inducible expression and activity of CSE in MPC and articular chondrocytes
To identify potential pathways regulating cytokine-induced H2S synthesis, chondrocytes were incubated with TNF-␣, IL-1␤ and IL-6 in the presence of various inhibitors of p38, ERK1/2, JNK and NF-B; pathways strongly implicated in chronic inflammatory disease [47][48][49][50]. Pharmacological inhibition of p38, NF-B (Fig. 2C) and ERK1/2 but not JNK (Fig. 2D) significantly inhibited cytokine-and LPS-induced H2S synthesis in chondrocytes. To investigate whether the compounds used to inhibit p38, ERK1/2, JNK and NF-B could also inhibit the CSE activity, we incubated these compounds with cell lysates from LPS-treated chondrocytes. Figure 2E shows that the compounds used in our study did not significantly lower H2S synthesis, for example the compounds themselves did not inhibit CSE/CBS activity directly. To confirm the molecular requirement for NF-B in cytokine-induced H2S synthesis, we performed additional experiments using NF-B p65-siRNA. Figure 2F shows that in MPCs, p65 NF-B siRNA treatment, but not non-coding controls (NCC-1 and NCC-2), significantly lowered cytokine-induced H2S synthesis, confirming the requirement for NF-B activation in mediating cytokine-inducible synthesis of H2S.

Effects of endogenous and pharmacological H 2 S on oxidative/nitrosative stress-induced cell death
We initially investigated the effects of the commonly used sulfide salts, sodium sulfide (Na2S) and sodium hydrosulfide (NaSH), and the slow-release H2S donor compound GYY4137 on cell viability using MTT and LDH release assays. Na2S, and to a lesser extent NaSH, induced significant concentration-dependent cytotoxicity in MPCs (Fig. 3A and B) and in alginate cultured chondrocytes ( Fig. 3C and D). These observations are consistent with the effects of Na2S and NaSH on other primary cell monolay-ers [51][52][53][54]. In contrast, no significant cytotoxicity was observed in MPCs and chondrocytes exposed to GYY4137 over the same concentration range (Fig. 3). This observation is also consistent with the findings of others [34,55]. Therefore future experiments using Na2S and NaSH were precluded due to overt cytotoxicity. Preliminary control experiments showed that in the cell culture media in the absence of cells, Na2S, NaSH or GYY4137 treatment alone (up to 500 mol/l) did not significantly reduce MTT to formazan or inhibit LDH activity (data not shown), for example, the H2S donors alone did not interfere with the viability assays.
Further control experiments showed that the addition of 'decomposed' GYY4137 did not significantly reduce cell viability (MTT assay; mean Ϯ S.D. 98.5 Ϯ 3.2%; cf. control, 102.1 Ϯ 3.1%) or induce significant LDH release (mean Ϯ S.D. 2.3 Ϯ 1.2%; cf. control, 3.4 Ϯ 2.1%). Figure 4 shows that, in agreement with previous studies using chondrocytes, exposure of MPC cells and chondrocytes to the peroxynitrite donor SIN-1 (100 and 500 mol/l), H2O2 (100 and 500 mol/l) and the 4-HNE (20 and 50 mol/l) induced significant cell death as determined by MTT (Fig. 4A) and LDH release (Fig. 4B) assays. These concentrations and treatment conditions were identified from the literature and from preliminary investigations by us. These treatment conditions were then employed in subsequent experiments to determine whether endogenous and/or pharmacological H2S could be cytoprotective.
Oxidative stress-induced cell death in MPCs ( Fig. 4C and D) was significantly inhibited by treatment of cells with L-cysteine. However, this protective effect was significantly removed in the presence of the CSE inhibitor PAG and by the CBS inhibitor AOAA. Furthermore, oxidative stress treatment in the presence of PAG and AOAA alone significantly increased cell death in H2O2-, SIN-1and 4-HNE-treated cells collectively, suggesting that endogenous H2S could protect against cell injury. Because CSE-and CBSgenerated H2S occur in a slow and steady manner [11,56,57], we next investigated whether the slow release H2S donor compound GYY4137 could prevent oxidative stress-induced cell death. Figure 4D shows GYY4137 preserved cellular viability against all three oxidative insults in a concentration-dependent manner. This effect was not observed with 'decomposed' GYY4137 added as a control. To confirm the molecular requirement for CSE-and CBSderived H2S in mediating cytoprotection, we next examined the effects of siRNA-mediated CSE and CBS protein knockdown in chondrocytes. In CSE-and CBS-siRNA-treated cells the cytoprotective effect of L-cysteine against oxidative injury was significantly reduced but preserved in cells transfected with two noncoding siRNA ( Fig. 5A and B). Small but statistically significant increases in cell death induced by SIN-1 (Fig. 5A), H2O2 (Fig. 5B) and 4-HNE (Fig. 5C) were observed with CSE and CBS-siRNA treatment, in agreement with the effects of PAG and AOAA in MPCs ( Fig. 4C and D). Pharmacological H2S, supplied via GYY4137, further significantly inhibited oxidative stress-induced cell death when added to CSE-treated and CBS-siRNA-treated cells, that is, the lack of enzymatically generated H2S could be overcome by pharmacological H2S.

Effects of pharmacological H 2 S on pro-cell survival signalling
To determine the potential pathways mediating H2S-mediated cytoprotection we next investigated the effects of GYY4137 on Akt and ERK1/2 activation. Figure 6A shows that GYY4137, but not the 'decomposed' donor, induced a time-and concentrationdependent increase in the phosphorylation (Thr 308 ) of the pro-cell survival protein Akt in chondrocytes. Figure 6B shows that L-cysteine treatment of MPCs for 6 hrs significantly increased the phosphorylation of Akt whereas PAG and AOAA (added as inhibitors of CSE and CBS, respectively) significantly lowered L-cysteine-induced Akt phosphorylation. Figure 6C shows that incubation of cells in the presence of GYY4137 (0-500 mol/l) and L-cysteine (1 mmol/l) for 6 hrs also significantly increased ERK1/2 phosphorylation.
To determine whether Atk was required for H2S-mediated cytoprotection, MPCs and chondrocytes were treated with pharmacological inhibitors of Akt signalling in the presence of GYY4137 and cell death induced by oxidant stress determined as described earlier. Figure 7 shows that pharmacological inhibition of Akt and upstream PI-3 kinase and to a lesser extent ERK1/2, significantly reduced the ability of GYY4137  Fig. 7C) suggesting activation of Akt and ERK1/2 signalling pathways was required for H2S-mediated cytoprotection. Furthermore, siRNA-mediated Akt protein knockdown in chondrocytes showed that activation of Akt was required to confer cytoprotection by GYY4137 against oxidative stress-mediated cell death (Fig. 7D).

Effects of H 2 S on oxidative stress-mediated mitochondrial dysfunction
There is increasing evidence for perturbed mitochondrial function in inflammatory and degenerative joint disease [29,43,58] and that mitochondrial dysfunction may potentiate inflammation [29]. Because SIN-1 [26], 4-HNE [42] and H2O2 [59] have previously been shown to collapse mitochondrial ⌬⌿m in human chondrocytes, we next investigated whether pharmacological H2S could protect mitochondria in situ from oxidative injury by examining mitochondrial ⌬⌿m, mitochondrial ATP and cytoplasmic cytochrome c. In MPCs, inhibition of CSE (with PAG) or CBS (with AOAA) significantly increased the loss of mitochondrial ⌬⌿m induced by SIN-1 (Fig. 8A), H2O2 and 4-HNE (Fig. 8B), whereas GYY4137 added at a final concentration of 200 mol/l or 500 mol/l significantly inhibited mitochondrial toxicity. In agreement with these observations, treatment of chondrocytes with Fig. 4 Inhibition of oxidative stress-induced cell death by endogenous and pharmacological H2S. MPCs and chondrocytes were exposed to the peroxynitrite donor SIN-1 and H2O2 (100 mol/l and 500 mol/l) and the lipid peroxidation product 4-HNE (20 mol/l and 50 mol/l) and cell death determined by MTT (A) and LDH release (B) assays. (C) Effect of endogenously generated H2S on oxidative stress-induced cell death. MPCs were exposed to H2O2 (200 mol/l), SIN-1 (500 mol/l) and 4-HNE (25 mol/l) for 18 hrs in the presence and absence of CSE and CBS inhibitors PAG and AOAA (1 mmol/l), respectively. L-cysteine (1 mmol/l) was added as a substrate for CSE and CBS. Cell death was determined by MTT assay. (D) Effect of GYY4137 and 'decomposed' GYY4137 on oxidative stress-induced cell death. MPCs were exposed to oxidative insult in the presence and absence of the slow release H2S donor GYY4137 (200 mol/l and 500 mol/l) or decomposed GYY4137 and cell death determined by MTT assay as described in Materials and Methods. Data are expressed as mean Ϯ S.D. of six determinations. *P Ͻ 0.05, **P Ͻ 0.01, ***P Ͻ 0.001, cf. oxidant treatment alone. CSE-siRNA or CBS-siRNA but not non-coding controls (see figure legend) similarly increased oxidative stress-induced collapse of mitochondrial ⌬⌿m. In these additional experiments, GYY4137 (500 mol/l) also inhibited SIN-1-, H2O2and 4-HNE-induced mitochondrial toxicity in CSE-siRNA and CBS-siRNA-treated chondrocytes (Fig. 8C).

Discussion
Hydrogen sulfide is emerging as an important mediator in acute and chronic inflammation [3] as well as vascular pathologies such as hypertension, obesity, diabetes and kidney disease (reviewed in [60]). However, the role of H2S in human inflammatory joint disease, whether human joint cells synthesize H2S and how joint cells respond to H2S has not been investigated in detail. Previous indirect evidence has suggested perturbed H2S synthesis in RA. For example, serum levels of the CSE/CBS substrates L-homocysteine, L-cysteine and L-cystathionine are elevated [61,62], and erythrocyte levels of S-thiolmethyltransferase (TMT), an enzyme potentially capable of 'detoxifying' H2S in vivo are decreased in RA plasma compared to healthy controls [63]. More recently H2S has been shown to be present in knee-joint SF aspirates from RA and OA patients [9] and other arthritides [3]. Consistent with a role of H2S in mediating inflammatory signalling, significantly higher H2S levels were recently observed in RA compared to OA SF and levels significantly correlated with inflammatory scores, including neutrophil and total white cell count [9], but the precise source of H2S in SF has not been determined. Furthermore, in an in vivo murine model of acute arthritis induced by kaolin/carrageenan [12], 30-50 M of Na2S inhibited leucocyte adhesion in postcapillary venules in acutely inflamed mouse knees. In sharp 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 pro-inflammatory vasodilatory mediators such as PGE2, • NO and histamine. Activated rodent neutrophils, macrophages and vascular endothelial cells have been shown to synthesize H2S in response to pro-inflammatory stimuli and if equivalent human cells are present in the inflamed RA joint they could also generate H2S. This study also suggests that resident joint cells, when exposed to a strongly pro-inflammatory milieu, would also synthesize H2S. We were unable to compare the effects of Na2S and NaSH with the previous studies on rodent cell lines or immortalized chondrocyte monolayers using these compounds as Na2S and NaSH induced significant cytotoxicity in MPCs and chondrocytes (Fig. 3). These findings are in agreement with studies on other primary cells such as acinar cells, pancreatic ␤-cells and vascular smooth muscle cells and could highlight important differences in responsiveness to H2S between primary cells and cell lines.
Furthermore, NaSH (and Na2S) may not be ideal compounds for studying the physiology (or pathophysiology) of H2S because these sulfide salts generate an instantaneous and very short lived (Ͻ5 sec.) bolus of concentrated H2S (as well as Na ϩ ), whereas enzymatic CSE and CBS-derived H2S synthesis is considerably slower over a much longer time period [56,57,64]. The manner in which cells and tissues are exposed to H2S may influence the cellular or tissue response [11]. To this end, we used GYY4137 to generate 'pharmacological' H2S and L-cysteine to generate 'endogenous' H2S to more accurately study the role of physiologically generated H2S.
Our study also suggests that the control of H2S synthesis through the induced activity of CSE proceeds through well-defined pathways strongly associated with the regulation of inflammatory signalling, NF-B, p38 and ERK1/2. The pathways controlling H2S synthesis have largely been overlooked and the published studies to date have only focused on the effects of added H2S (invariably as Na2S or NaSH) on cells and tissues. The H2S donor GYY4137 has recently been shown to generate H2S in a manner comparable to enzymatically synthesized H2S from CSE/CBS and does not induce significant cytotoxicity even at mmol/l concentrations [11,34]. GYY4137 has also been shown to inhibit the synthesis of pro-inflammatory mediators (e.g. TNF-␣, IL-6, IL-1␤, PGE2 and • NO) in LPS-stimulated murine macrophages [11] and IL-8 secretion by primary human pulmonary airway smooth muscle cells in vitro [65]. Furthermore, in a rat model of endotoxemia GYY4137 decreased the plasma levels of these inflammatory mediators as well as inhibiting their synthesis in isolated whole blood, limiting endotoxin-induced tissue damage, oedema and inflammation [33]. It is therefore possible that the up-regulation of H2S synthesis, previously observed in RA SF, may represent a novel endogenous mechanism for limiting inflammation in the joint.
There is strong evidence for a contribution of oxidative stress in mediating the various aspects of inflammatory and degenerative joint disease, for example, cartilage destruction, mitochondrial dysfunction, cell death of resident joint cells and pro-inflammatory signalling (reviewed in [39]). This study suggests an additional role for H2S in human joint cells: cytoprotection against oxidative injury. Is it therefore possible that elevated H2S synthesis in the inflamed joint counteracts increased oxidative/nitrosative stress? In MPCs and chondrocytes oxidative stress-induced mitochondrial dysfunction and cell death was significantly inhibited when cells were exposed to GYY4137 within the concentration range of 100-500 mol/l (but not its decomposed control) or L-cysteine. The ability of L-cysteine to protect against cell death was inhibited by the treatment of cells with CSE and CBS inhibitors (PAG and AOAA, respectively) and by siRNA-mediated protein knockdown. PAG and AOAA have been used extensively to reduce cellular and tissue synthesis of H2S in a variety of cell and whole animal model systems, but these compounds target the PLP binding site of CSE and CBS and may also inhibit other PLP-dependent enzymes. There are currently no completely selective inhibitors of CSE-and CBS-derived H2S synthesis (reviewed in [3]) necessitating the requirement for complementary and confirmatory experiments using CSE and CBS siRNA.
Recently, in simple in vitro and cell free assays, Na2S and NaSH have been shown to 'scavenge' pro-inflammatory oxidants such as • NO [66], ONOO Ϫ [67], OCl Ϫ [68,69] H2O2 [70] and superoxide (O2 •Ϫ ) [71] as well as 'destroy' lipid peroxides [72,73] and inhibit atherogenic modification of low density lipoprotein [74]. However, more recent in vitro kinetic studies have suggested that the rate constants for the reaction of NaSH-generated H2S with the above oxidants are not sufficiently high enough for oxidant 'scavenging' to solely account for the cytoprotective effects of H2S on cells, suggesting additional cellular mechanisms must be involved [75]. In support of this, NaSH and Na2S have been shown to up-regulate glutathione synthesis via ␥-glutamylcysteine synthetase, to increase cysteine uptake in neuronal cells [76,77] and to activate Nrf-2 signalling, thereby conferring cytoprotection in cardiac cells and tissues [78,79] and preserving mitochondrial integrity. However, the mechanisms for these observations are unclear. In this study, slow release of H2S by GYY4137 or stimulation of endogenous H2S production with L-cysteine induced significant time-and concentration-dependent phosphorylation of Akt and ERK1/2 and prevented the loss of mitochondrial ⌬⌿m and cell death induced by oxidant species that have been well characterized to have detrimental effects on the extracellular matrix and cartilage producing cells. Furthermore, inhibitors of PI3K, Akt and ERK significantly reduced the protective effect of GYY437. The protective effects of L-cysteine were significantly reduced in the presence of PAG and AOAA and nearly absent in CSE-or CBS-siRNA-treated cells. This suggests that a potential role of endogenous H2S is to activate PI3K-Akt/ERK cytoprotective pathways and preserve mitochondrial and cellular integrity and function. Recently H2S, albeit derived from NaSH, was shown to decrease IL-6 and IL-8 synthesis in the immortalized human C29/I2 chondrocyte cell line [80] grown in monolayer culture suggesting the possibility that H2S could also regulate the inflammatory response in cartilage cells, in agreement with previously published studies in other cell types.
In summary, our study shows for the first time that (1) chondrogenically differentiated MPCs and HACs synthesize H2S via CSE and CBS, (2) pro-inflammatory cytokines induce CSE expression and activity via p38-ERK-NF-B-dependent pathways, (3) endogenous and pharmacological H2S inhibit oxidant-induced mitochondrial dysfunction and cell death through pathways involving Akt/PI3K-dependent signalling and (4) induced H2S synthesis may represent a novel endogenous mechanism to limit cartilage destruction, cell death and inflammation in the joint. Therefore, the role of H2S in chronic joint inflammation deserves further attention. Controlling H2S synthesis may represent a novel opportunity for therapeutic intervention in human chronic inflammatory diseases such as rheumatoid arthritis.