The reaction of hydrogen sulfide (H2S) with peroxynitrite (a key mediator in numerous pathological states) was studied in vitro and in different cellular models. The results show that H2S can scavenge peroxynitrite with a corresponding second order rate constant of 3.3±0.4×103 M−1·s−1 at 23°C (8±2×103 M−1·s−1 at 37°C). Activation parameters for the reaction (ΔH‡, ΔS‡ and ΔV‡) revealed that the mechanism is rather associative than multi-step free-radical as expected for other thiols. This is in agreement with a primary formation of a new reaction product characterized by spectral and computational studies as HSNO2 (thionitrate), predominantly present as sulfinyl nitrite, HS(O)NO. This is the first time a thionitrate has been shown to be generated under biologically relevant conditions. The potential of HS(O)NO to serve as a NO donor in a pH-dependent manner and its ability to release NO inside the cells has been demonstrated. Thus sulfide modulates the chemistry and biological effects of peroxynitrite by its scavenging and formation of a new chemical entity (HSNO2) with the potential to release NO, suppressing the pro-apoptotic, oxidative and nitrative properties of peroxynitrite. Physiological concentrations of H2S abrogated peroxynitrite-induced cell damage as demonstrated by the: (i) inhibition of apoptosis and necrosis caused by peroxynitrite; (ii) prevention of protein nitration; and (iii) inhibition of PARP-1 [poly(ADP-ribose) polymerase 1] activation in cellular models, implying that a major part of the cytoprotective effects of hydrogen sulfide may be mediated by modulation of peroxynitrite chemistry, in particular under inflammatory conditions.
The past decade has witnessed an increasing body of evidence that hydrogen sulfide (H2S) plays important roles in both physiology and pathology [1,2]. Produced in cells by the action of two differently expressed enzymes, cystathionine γ-lyase and cystathionine β-synthase , H2S predominantly regulates cardiovascular  and nervous systems . Alongside its physiological roles, the beneficial effects of H2S-based therapy on various animal disease models have been documented as well . However, all the extensive reviews that deal with H2S underline the fact that chemical evidence and explanations for its biological effects are still elusive [1–3,6].
As a gaseous molecule, H2S easily passes through cell membranes by simple diffusion . With pKa1 of ~7.0 and pKa2>14, there is essentially no S2− in biological tissues, whereas the ratio of H2S/HS− varies with the temperature and pH . The actual concentration of H2S in vivo is still a matter of debate (ranging from undetectable to high micromolar). However, the latest results of Shen et al.  clearly demonstrated a low micromolar concentration range of H2S in plasma. The concentration of HS− in the cells is confirmed to be much higher [1,2].
Interestingly, most of the disease models where H2S therapy showed beneficial effects relate to inflammation, inflammation-based diseases  or ischaemia–reperfusion injury [11,12], all of which are hallmarked by increased production of reactive oxygen/nitrogen species, in particular peroxynitrite (ONOO−/ONOOH). Peroxynitrite (ONOO−) and its protonated form (ONOOH) generated in the diffusion-controlled reaction between NO and superoxide (O2•−) (~1×1010 M−1·s−1) , in particular by stimulated macrophages , are implicated in the development of a wide variety of pathological conditions, including inflammation, ischaemia–reperfusion and neurodegenerative disorders among others [15,16]. Peroxynitrite reacts with important biomolecules such as DNA, proteins, lipids and sugars, leading to cell damage and eventually to cell death via apoptosis and/or necrosis [15,16]. Although the actions of peroxynitrite in pathologies are numerous, a mechanism that has received the greatest attention is its ability to nitrate tyrosine residues in proteins [15,16]. Nitration of MnSOD (manganese superoxide dismutase), for example, has been reported to be particularly specific for ischaemia–reperfusion injury .
The interplay between these two small inorganic species with generally opposing activities, pro-oxidative in the case of ONOO− against pro-reducing activity of H2S, is not only chemically intriguing, but can also be involved in molecular events behind the beneficial effects of H2S-based therapy. Indeed, Whiteman et al.  showed that H2S prevents nitration of neuronal cells when the cell culture was exposed to synthetic peroxynitrite. Recently Nagy and Winterbourn  reported that H2S reacts with another neutrophil-derived mediator of inflammation, HOCl, at a diffusion-controlled rate (2×109 M−1·s−1). However, Carballal et al.  published a kinetic study on the reaction of peroxynitrite (and other biologically relevant oxidants, including HOCl) with H2S using single wavelength UV–vis (UV–visible) stopped-flow measurements, proposing that the mechanism of the reaction is the same as with other thiols and that H2S could have little, if any, physiological relevance when compared with, for example, glutathione.
In the present study we re-address this reaction in more detail by applying time-resolved (rapid-scan) and high-pressure stopped-flow UV–vis measurements, electrochemical (amperometry and cyclic voltammetry) and MS methods. Our kinetic and computational studies show that the proposed and widely used mechanism to describe the reaction of peroxynitrite with thiols is not operative for the reaction with H2S. Furthermore, we identify in the present study the formation of a previously unidentified reaction product with NO-donating properties, having a predominant sulfinyl nitrite character. We also demonstrate that H2S efficiently prevents MnSOD nitration in vitro, preventing general protein nitration as well as peroxynitrite-induced necrosis and apoptosis in cell cultures when used at pharmacological concentrations, which could partially explain the beneficial effects of H2S-based therapy on ischaemia–reperfusion injury and other disease states where the formation of peroxynitrite is pivotal.
Unless stated otherwise, all chemicals were purchased from Sigma–Aldrich. KPi (phosphate buffer; 300 mM, pH 7.4) was prepared with nano-pure water, stirred with Chelex-100 resins to remove traces of heavy metals and kept above the resins until used. Na2S was purchased in its anhydrous form, opened and stored in glove box (<1 p.p.m. O2 and <1 p.p.m. H2O). Stock solutions (1 M and 1 mM) of sodium sulfide were prepared in the glove box using argon-bubbled nano-pure water and stored in glass vials with PTFE [poly(tetrafluoroethylene)] septa at +4°C, for a maximum of 1 week. The concentration of H2S was determined using a H2S-selective electrode (World Precision Instruments). Gas-tight Hamilton syringes were used throughout the study. Peroxynitrite was prepared as described previously . Hydrogen peroxide is eliminated by passing the peroxynitrite solution through a manganese dioxide column and total removal was confirmed by a H2O2-sensitive electrode (World Precision Instruments). Nitrite contamination of the peroxynitrite solution was less than 2% as determined using the Griess method. Peroxynitrite solutions were stored at −25°C. Thawed solutions were not re-used. The same procedure (on a smaller scale) was used for preparation of 15N-labelled ONOO−. Recombinant chicken Annexin A5 was obtained from Responsif and labelled with FITC employing standard methods.
UV–vis spectrometric studies
All spectrophotometric studies were done on a HP 8452A diode array spectrophotometer connected to a Dell computer equipped with Olis SpectralWorks software. Anaerobic measurements were performed in anaerobic cuvettes.
Kinetic data were obtained by recording time-resolved UV–vis spectra using a modified μSFM-20 Bio-Logic stopped-flow module combined with a Huber CC90 cryostat and equipped with a J&M TIDAS high-speed diode array spectrometer with combined deuterium and tungsten lamps (200–1015 nm wavelength range). Isolast O-rings were used for all sealing purposes, and solutions were delivered with 10 ml gas-tight Hamilton syringes. The syringes were controlled by separate drives, allowing the variation of the ratio of mixing volumes used in the kinetic runs. Data were analysed using the integrated Bio-Kine software version 4.23 and also the Specfit/32™ program. At least seven kinetic runs were recorded for all conditions, and the reported rate constants represent the mean values. The stopped-flow instrument was thermostated to the desired temperature ±0.1°C. All kinetic measurements were carried out under pseudo-first-order conditions, i.e. H2S concentration was in excess (peroxynitrite concentration was 1.4×10−4 M). The reactions were studied at an ionic strength of 0.3 M (KPi pH 7.4). The desired concentration of H2S was prepared by injecting a defined volume of sodium sulfide stock solution into argon-purged 300 mM KPi, pH 7.4, prior to measurement. Spontaneous decomposition of peroxynitrite was measured by mixing 140 μM ONOO− in 20 mM KOH with 300 mM KPi, pH 7.4. For the kinetic measurements under anaerobic conditions, solutions of H2S and peroxynitrite were bubbled with argon and prepared in a glove box in gas-tight Hamilton syringes. The anaerobic measurements were performed at room temperature (23±1°C).
Measurements under high pressure (10–135 mPa) were carried out using a custom-made high-pressure stopped-flow instrument, for which Isolast O-rings were used for all syringe seals . The stopped-flow instrument was thermostated at 10°C. The reactions were monitored at 310 nm. Values of ΔV‡ were calculated from the slope of plots of ln(k) against pressure in the usual manner based on eqn (1):
where R is the ideal gas constant and T is the thermodynamic temperature. All kinetic measurements were carried out under pseudo-first-order conditions. Sodium sulfide solution was added to 300 mM KPi, pH 7.4, prior to the measurements to give a concentration of 4 mM (2 mM after mixing in the observation cuvettes). The other solution contained 280 μM ONOO− (140 μM after mixing in the observation cuvettes) in 20 mM KOH. H2S concentration was selected on the basis of concentration-dependent measurements, which resulted in a linear dependence of kobs on the H2S concentration. The effect of pressure on the spontaneous decomposition of peroxynitrite was followed by mixing the solution of ONOO− with 300 mM KPi, pH 7.4.
Analysis of NO, H2S and O2
The fate of H2S, oxygen and NO during the course of the reaction was monitored by a Free Radical Analyzer (World Precision Instruments) connected to a Dell computer equipped with DataTrax software for the signal processing. Experiments were performed in a four channel chamber (World Precision Instruments) with all three electrodes at the same time or each of them separately. KPi (2 ml of 50 mM, pH 7.4) was added into the chamber and the electrodes were immersed into it. Depending on the type of measurement, different concentrations of sodium sulfide solution were injected followed by the addition of peroxynitrite solution. For the anaerobic measurements, the chamber was closed with a gas-tight plunger and purged with argon for 20 min. When pH-dependent NO release from the product was monitored, a reaction mixture containing 1 mM peroxynitrite and 1 mM H2S in 300 mM KPi, pH 7.4, was injected into the chamber equipped with a NO-sensitive electrode and filled with 50 mM KPi, pH 6.2, or 50 mM acetate buffer, pH 5.1.
Cyclic voltammetric measurements were carried out using an Autolab instrument with a PGSTAT 30 potentiostat. A conventional three electrode arrangement was employed consisting of a gold working disk electrode (Metrohm, geometric area: 0.07 cm2), a platinum wire (Metrohm) as the auxiliary electrode and Ag/AgCl as the reference electrode. The solution for the measurements contained 1 mM sodium sulfide in 300 mM KPi, 1 mM peroxynitrite in 300 mM KPi, and a solution that was made adding both sulfide and peroxinitrite solutions into 300 mM KPi, pH 7.4. Solutions were thoroughly degassed with nitrogen prior to starting the experiments. All experiments were performed at room temperature and under nitrogen.
ESI–MS (electrospray ionization–MS)
Using a syringe pump at a flow rate of 240 ml per h, the aqueous solutions were infused into an orthogonal ESI source of an esquire6000 ion trap mass spectrometer (Bruker). Nitrogen was used as the nebulizing gas at a pressure of 10 psi (1 psi=6.9 kPa) and as the drying gas at a temperature of 280°C with a flow rate of 5 l per min. The ion trap was optimized for the respective target mass of the ions under investigation. The source voltages varied with this optimization. All experiments were carried out in the positive-ion mode. The samples were prepared by mixing 0.1 mM (final concentration) peroxynitrite or 15N-labelled peroxynitrite with 0.1 mM (final concentration) Na2S in 20 mM KPi, pH 7.4.
MnSOD (20 μM) from Escherichia coli was exposed to 25 or 50 μM single doses of synthetic peroxynitrite in 50 mM KPi, pH 7.4, in the presence and absence of 50 μM H2S. The samples were analysed on a maXis, a high-resolution ESI–TOF (time-of-flight) mass spectrometer (Bruker). The samples were injected using a syringe pump at a flow rate of 240 ml per h. Nitrogen was used as the nebulizing gas at a pressure of 10 psi and as the drying gas at a temperature of 180°C and a flow rate of 5 l per min. All experiments were carried out in the negative-ion mode and obtained spectra were deconvoluted and further processed using Data Analysis software provided by Bruker Daltonics. Both native and peroxynitrite-modified MnSOD were dialysed extensively against water before MS studies. Analytical samples were prepared by mixing native protein samples in water with a mixture of 0.1% formic acid in water/acetonitrile (1:1, v/v).
Oxidation of DHR (dihydrorhodamine-123)
The oxidation of non-fluorescent DHR to fluorescent rhodamine-123 was measured using a Fluorolog 3-22 spectrofluorimetar (Jobin Yvon), with λexcitation=503 nm and λemission=526 nm. The formation of rhodamine-123 was also quantified spectrophotometrically at 500 nm (ϵ=78000 M−1·cm−1).
Cell culture experiments
The T-lymphoblastic Jurkat cell line was obtained from the A.T.C.C. and was cultured in 96-well plates at 37°C in a humidified atmosphere in the presence of 5.5% CO2 in culture medium (RPMI 1640) supplemented with 10% heat-inactivated FBS (fetal bovine serum), 1% glutamine, 1% Hepes and 1% penicillin/streptomycin (all from Gibco).
HeLa cells were obtained from ECACC (European Collection of Animal Cell Cultures). Cells were plated in 35-mm-diameter μ-dishes (ibidi) with DMEM (Dulbecco's modified Eagle's medium) containing 10% FBS, 2 mM glutamine and 1% penicillin/streptomycin at 37°C and 5% CO2.
Detection of apoptosis, necrosis and cell viability in Jurkat cells
Oxidative stress in Jurkat cells was induced by adding 100 μM SIN-1 (3-morpholinosydnonimine hydrochloride), which produces a flux of 1.2 μM/min of peroxynitrite to the cell culture (100000 cell/200 μl). After 2 h, 100 μl of the culture was harvested for viability measurement and 100 μl fresh medium was added. Cell viability measurements were repeated 24 h and 48 h after treatment. To detect apoptosis and necrosis, cells were analysed by flow cytometry for phosphatidylserine surface exposure and membrane integrity by staining in Ringer's solution with Annexin A5-FITC (400 ng/ml) and the supravital dye PI (propidium iodide; 1 μg/ml). The percentages of cell viability (annexin A5-negative and PI-negative), apoptosis (annexin A5-positive and PI-negative) and necrosis (double positive cells) were recorded.
Intracellular detection of NO with DAF-FM (4-amino-5-methylamino-2′,7′-difluororescein diacetate)
For the intracellular detection of NO, cells were incubated with 30 μM DAF-FM for 10 min. The dye was washed out three times, cells were placed into the medium and treated for 30 min with 100 μM peroxynitrite, 100 μM decomposed peroxynitrite, 100 μM sodium sulfide or a combination of both and 100 μM HSNO2 made by mixing equimolar concentrations of sulfide and peroxynitrite prior to addition. Fluorescent microscopy was carried out using a Carl Zeiss Axiovert 40 CLF inverted microscope equipped with a green fluorescence filter (λexcitation=450–490 nm, λemission=500–700 mn) and Axiocam Icm1. All experiments were performed in triplicate. Images were processed using ImageJ software where semi-quantitative fluorescence intensity was determined.
Detection of apoptosis in HeLa cells
Apoptosis in HeLa cells was quantified using methods described previously . Briefly, cells were stained with Hoechst 33342 DNA dye after treatment with 100 μM peroxynitrite, 100 μM Na2S or with both compounds, and the nuclear morphology was examined. Uniformly stained nuclei were scored as healthy viable cells, whereas condensed or fragmented nuclei were marked as apoptotic. Experiments were performed in triplicate. For presentation, photomicrographs were post-processed in Adobe Photoshop where cells were artificially coloured blue (for healthy cells) or red (for those marked as apoptotic). Parts of the nuclei that were on the borders of the micrographs were removed and not included in the total cell count.
Imunocytochemical detection of nitrotyrosine formation
To detect nitrotyrosine formation, HeLa cells were exposed to 100 μM sodium sulfide, 100 μM SIN-1, 100 μM peroxynitrite, a combination of 100 μM SIN-1 and sulfide or a combination of 100 μM peroxynitrite and sulfide over 1 h. Cells were fixed in 4% para-formaldehyde for 10 min at room temperature and washed three times with PBS. Non-specific binding was blocked using 1% BSA containing 0.3% Tween 20 for 2 h at room temperature. Incubation with primary monoclonal anti-(3-nitrotyrosine) antibody produced in mouse (Sigma–Aldrich) was performed overnight at 4°C according to the manufacturer's instructions. Cells were washed three times with PBS and incubated with FITC-conjugated secondary anti-(mouse IgG) produced in goat (Sigma–Aldrich) at room temperature for 2 h.
ABC (ATP-binding cassette) transporter functional studies
PBMCs (peripheral blood mononuclear cells) and granulocytes were isolated from heparinized peripheral blood of normal healthy volunteers by standard density centrifugation. Informed consent was obtained from all blood donors, the study received final approval from the ethics committee of the University Hospital Erlangen and was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association. Fluorochrome loading was performed by incubation of 1×107 cells/ml in PBS containing 2 mM CFSE (carboxyfluorescein diacetate/succinimide ester; Molecular Probes) for 30 min at 37°C. Further dye uptake was prevented by washes in ice-cold PBS. To assess the functional dye export mediated by ATP-dependent membrane transporters, cells were cultured at 37°C. The time course of the cellular fluorochrome export was monitored by two-colour flow cytofluorometry using an EPICS-XLt flow cytofluorometer (Coulter). In all experiments, FS (forward scatter) against SS (side scatter) dot plots were used to strictly gate on viable cells. This gate contained less than 5% of annexin V-FITC-binding cells (Roche Biochemicals) and below 1% of PI-permeable cells. To modulate the export of fluorochromes, cells were treated with 50 μM sodium sulfide, 150 μM SIN-1 or a combination of both.
Geometries were optimized at the OLYP/6–31+G(d), B3PW91/6–31+G(d), B3PW91/6–311+G(d,p) and MP2/6–31+G(3df,2p) levels of theory including frequency analyses to disclose the nature of the stationary points (as implemented in the Gaussian 09 program package [24,25]). UV–vis calculations were performed employing TD-B3PW91/6–31+G(3df,2p) method.
Definitions and choice of reaction conditions
In the current literature that deals with the physiological effects of H2S, all three species, H2S, HS− and S2−, are commonly referred to as H2S. In the present study, all experiments were performed at pH 7.4, and for clarity the H2S/HS− equilibrium mixture is referred to as ‘H2S’ unless stated otherwise. Also, the pKa of ONOOH is 6.8 and both the acid and the anionic form are present at physiological conditions (pH 7.4) . Throughout the present paper, the term peroxynitrite refers to both anion and acid. Furthermore, the current view on the mechanism of peroxynitrite decomposition is divided, with evidence for both partial homolytic generation of NO2 and OH radicals [26,27] and bimolecular reaction . However, a previous study showed that, when used at concentrations higher than 150 μM and at a physiological pH, the bimolecular reaction takes place with concomitant formation of O2NOOH . In order to avoid this, the concentration of peroxynitrite used in this study was <150 μM.
We determined the rate constant for decomposition of peroxynitrite at pH 7.4 in the absence and the presence of ‘H2S’ by stopped-flow time-resolved UV–vis measurements. In the presence of H2S (750 μM–3 mM), decomposition of peroxynitrite was significantly faster and the obtained second order rate constant was 3.3±0.4×103 M−1·s−1 at 23°C (8±2×103 M−1·s−1 at 37°C) (Figure 1A). Furthermore, detection of a new peak at 408 nm corresponding to the yellow reaction product (Figure 1B) was observed. The kinetics of both the peroxynitrite decomposition and product formation excellently fit to a single-exponential function (Inset in Figure 1B and Supplementary Figure S1A at http://www.BiochemJ.org/bj/441/bj4410609add.htm). On the basis of the global analyses of time-resolved spectra, the spectrum of the reaction product contains two absorption maxima at 289 and 408 nm (Supplementary Figure S1B), which is in excellent agreement with the spectral characteristics of the product obtained experimentally (Figure 1C).
Stopped-flow measurements of the reaction between ‘H2S’ and peroxynitrite
In order to gain mechanistic insight into this reaction we performed a temperature- and pressure-dependent kinetic study, which allowed us to determine activation enthalpy (ΔH‡), activation entropy (ΔS‡) and volume of activation (ΔV‡) (Table 1). On the basis of the concentration dependence of kobs as a function of temperature (4–37°C) (Figure 1A), a good linear correlation between lnk (where k is the second order rate constant at a particular temperature) and T−1 was obtained (Figure 2A), and from the corresponding slope and intercept, ΔH‡ and ΔS‡ were calculated. High-pressure stopped-flow analysis of the reaction at 10°C revealed that the observed rate constant increases with the pressure increase (10–135 mPa) and the activation volume was calculated according to eqn (1) from the slope of the plot in Figure 2(B). Spontaneous decomposition of ONOOH was not greatly affected by pressure, giving ΔV‡=0.4±1 cm3·mol−1, which is in excellent agreement with the literature value of 1.7±1 cm3·mol−1 . A moderate value for ΔH‡ and ΔS‡ that is close to zero (Table 1) indicate that some sort of interchange mechanism with simultaneous bond formation and bond breaking operates in the rate-determining step, whereas the negative activation volume (−15.4±0.8 cm3·mol−1) implies an associative character of this process.
Temperature and pressure dependence of the reaction
H2S, O2 and NO on course of the reaction
In order to further analyse the reaction, we used a selective H2S sensor connected to a free radical analyser (World Precision Instruments) to follow ‘H2S’ consumption in the reaction. As shown in Figure 3(A), when injected into ‘H2S’ solution, peroxynitrite caused immediate removal of ‘H2S’ followed by a drop of current on the H2S sensor. The 25 μM solution of peroxynitrite completely removed 50 μM ‘H2S’ from the solution (not even detectable by the nose, which has a threshold of 0.5 p.p.b.), indicating that stoichiometry of the overall reaction is 2 moles of ‘H2S’ per 1 mole of peroxynitrite.
Consumption of H2S and oxygen over the course of the reaction
We also followed the reaction using an oxygen electrode. Peroxynitrite, when injected into a buffer solution, produces ~10% of O2  (see Supplementary Figure S2A at http://www.BiochemJ.org/bj/441/bj4410609add.htm). However, when ‘H2S’ was present in the solution, a rapid consumption of oxygen was observed (Figure 3B). ‘H2S’ spontaneously decomposes in oxygenated solution with consumption of O2, but consumption in this case is quite slow (Supplementary Figure S2A).
In order to test whether oxygen is a prerequisite for the reaction of ‘H2S’ with peroxynitrite, stopped-flow measurements under anaerobic conditions were performed. The obtained second-order rate constant was 3.8±0.5×103 M−1·s−1 at 23°C, which is equal (within the experimental error) to that found under aerobic conditions (Figures 1A and 3C), suggesting that oxygen does not play a role in the rate-determining step. Linear dependence of both aerobic and anaerobic kobs on ‘H2S’ concentration implies that the reaction is first-order regarding ‘H2S’ and not second-order as the stoichiometry may suggest. In order to check this, we measured H2S consumption upon addition of peroxynitrite into the argon-bubbled solution with an H2S-sensitive electrode. The results clearly indicate that in the absence of oxygen the overall stoichiometry is 1:1 (Figure 3D), suggesting that under aerobic conditions the second mole of ‘H2S’, together with O2, plays a role in the further transformation of the initial yellow product. Indeed, the yield of this product (as judged by the intensity of the peak at 407 nm) is higher under anaerobic conditions (Supplementary Figures S2B and S2C) and stable for 24 h, whereas under aerobic conditions the colour bleaches within 1 h. When the anaerobic solution was bubbled for 5 min with air, the peak at 407 nm completely disappeared after 30 min (Supplementary Figure S2D), but a new peak at 354 nm appeared that corresponds to nitrite.
Surprisingly, immediately after the addition of peroxynitrite into ‘H2S’ solution, NO was released from the solution as detected by a NO-sensitive electrode (Figure 4A). Furthermore, the NO-releasing properties of the reaction product were increased as the pH value of the solution decreased (Figure 4B).
The reaction product of peroxynitrite and ‘H2S’ releases NO
To get more information about the reaction product(s), the reaction mixture was analysed with ESI-ion trap MS and cyclic voltammetry. The reaction products of ‘H2S’ with both ONOOH and O15NOOH were analysed by ESI-ion trap MS. Since the buffer molarity was high, the intensities of the signals were relatively low (~104). However, the following peaks changed by m/z 1 when 15N-enriched ONOOH was used: m/z 80→m/z 81, m/z 96→m/z 97 and m/z 118→m/z 119 (Supplementary Figure S3A at http://www.BiochemJ.org/bj/441/bj4410609add.htm). The most intense peak was at m/z 127 and did not show a shift in the 15N-enriched sample. We assigned the peaks at m/z 81 and 119 as [HS15NO2+H]+ and [HS15NO2+K]+ respectively, and the peak at m/z 127 as [HSOK+K]+ (Figure 5A). Since the low mass limit of the instrument was m/z 50, it was not possible to observe MS2 spectra of the corresponding peaks.
Chemical characteristics of the reaction product
Redox behaviour of the product was also analysed by cyclic voltammetry. Decomposition products of ONOOH at pH 7.4 (~90% nitrate and ~10% nitrite, determined by the Greiss reaction and 15N NMR, results not shown) showed no oxidation or reduction waves (Supplementary Figure S3B), whereas the solution of ‘H2S’ showed one irreversible peak at −77 mV compared with Ag/AgCl (Supplementary Figure S3C). Cyclic voltammograms of the reaction mixture containing 1 mM Na2S and 1 mM ONOO− in 300 mM KPi, pH 7.4, showed an oxidation wave at +810 mV compared with Ag/AgCl (Figure 5B), confirming the generation of a new product with distinct redox properties.
To assign the UV–vis peaks of the observed yellow product, we performed calculations of UV–vis spectra of possible intermediates for the reaction between ONOO−+H2S/ONOOH+HS−. As possible products we considered compounds 1–3 (1). UV–vis calculations were performed at the TD-B3PW91/6–31+G(3df,2p) level. To verify applicability of this method, we also computed UV–vis spectra for peroxynitrite, for which the data exist. Optimization of compounds 1–3 (Scheme 1) with the DFT (density functional theory) and MP2 methods gave similar results in the case of compounds 1 and 2. However, for compound 3, MP2-optimized geometries are characterized with an elongated S–N bond [2.36 Å (1 Å=0.1 nm)] compared with DFT (2.04 Å) as shown on Figure 6. Thus UV–vis spectra calculations were performed for both geometries. The results are summarized in Supplementary Table S1 (at http://www.BiochemJ.org/bj/441/bj4410609add.htm). Only the structure of compound 3, HS(O)NO, showed the calculated absorbance maximum in the visible part of the spectrum, which was very close to the experimental one (417 nm calculated compared with 407 nm obtained experimentally).
Isomeric forms of the primary reaction product (HSNO2)
Optimized geometries of HS(O)NO with bond lengths in Å
Modulation of peroxynitrite reactivity by hydrogen sulfide
Oxidation of DHR by peroxynitrite has been used as a standard test for peroxynitrite oxidation/nitration capacity . As shown in Suplementary Figure S4(A), when 1 mM DHR was exposed to ~10 μM/min flux of peroxynitrite produced by SIN-1 (a peroxynitrite donor that produces equal fluxes of NO and O2•−) , its oxidation to rhodamine (monitored at 500 nm) was observed at pH 7.4 and 37°C. This effect was more pronounced in the presence of 5 mM HCO3−, an intracellularly present anion that is considered to be the primary physiological target of peroxynitrite . The presence of 50 μM H2S almost completely blocked rhodamine formation in the first 5 min in the absence of HCO3−, and slowed down the oxidation in the presence of bicarbonate, reducing the final oxidation yield to ~50%. Similar results were obtained when synthetic peroxynitrite was used. H2S showed concentration-dependant scavenging of peroxynitrite under both absence (Figure 7A) and presence of bicarbonate (Supplementary Figure S4B at http://www.BiochemJ.org/bj/441/bj4410609add.htm).
The effect of H2S on peroxynitrite-induced nitration of MnSOD and oxidation of DHR
We next aimed to detect the ability of ‘H2S’ to modify the nitration capacity of peroxynitrite. Nitration of protein tyrosine residues, in particular of MnSOD, has been used as a hallmark for peroxynitrite formation in vivo, especially in disease states related to inflammation and/or ischaemia–reperfusion injury and myocardial infarction . MnSOD (20 μM) from E. coli was exposed to 25 and 50 μM synthetic peroxynitrite in the presence and the absence of 50 μM H2S in 50 mM KPi, pH 7.4, that contained 1 mM bicarbonate. An increase in the absorbance at 430 nm was observed (Figure 7B), which was attributed to nitrotyrosine formation (due to the characteristic shift observed from 430 to 357 nm with decreasing pH) . Nitration yield displayed concentration dependence; however, it was completely abolished in the presence of ‘H2S’.
The samples were further analysed by negative ion mode ESI–TOF MS, and nitration of MnSOD treated with 50 μM peroxynitrite was confirmed by a deconvoluted mass of 22975 Da (m/z 1531.72 for a molecular ion) corresponding to the addition of one nitro group per protein subunit [Mr–H+NO2]− (molecular ion of native protein was m/z 1528.70 or 22930 Da after deconvolution). In the presence of ‘H2S’ no nitration was observed (Figure 7C).
Cellular consequences of H2S-induced modulation of peroxynitrite reactivity
In order to test whether the reaction of ‘H2S’ with peroxynitrite is efficient enough to compete with other biological targets, we performed several experiments on cultured cells. Peroxynitrite is known to induce both necrosis and apoptosis of the cells . In situ testing was performed on Jurkat cells exposed to 100 μM SIN-1A donor (1.2 μM/min flux of peroxynitrite) in the presence or the absence of 50 μM ‘H2S’ (this flux has been determined to be the normal flux of peroxynitrite production in the mitochondria of resting cells) . The effects on cell viability, necrosis and apoptosis were followed by cytofluorometry. As shown in Figure 8(A), protective effects of the 50 μM dose of ‘H2S’ were strong concerning both necrosis and apoptosis. ‘H2S’ itself did not show any effect on cell viability in the first 48 h after the treatment (results not shown). The deleterious effects of a smaller flux of ONOO− (0.3 μM/min) were efficiently scavenged even with 10 μM ‘H2S’ (results not shown).
Modulation of cytotoxic and pro-apoptotic effects of peroxynitrite by reaction with H2S
Furthermore, the repair of peroxynitrite-induced DNA damage is initiated by activation of PARP-1 [poly(ADP-ribose) polymerase 1] . A previous study showed that peroxynitrite-induced PARP-1 activation leads to deactivation of ABC transporters . In order to test the real potential of H2S to scavenge peroxynitrite in cells, cells were treated with 150 μM SIN-1 in the presence and absence of 50 μM sodium sulfide. As shown in Supplementary Figure S5 (at http://www.BiochemJ.org/bj/441/bj4410609add.htm), peroxynitrite inhibited pumping of the fluorochrome CFSE, whereas the treatment with ‘H2S’ partially restored the pumping activity.
Protective effects of ‘H2S’ were also tested on HeLa cells that were treated with synthetic peroxynitrite. Cell apoptosis was detected by Hoechst 33342 staining. Representative photomicrographs of nuclei morphology are shown on Figure 9(A) (and Supplementary Figure S6 at http://www.BiochemJ.org/bj/441/bj4410609add.htm). Treatment with peroxynitrite induced condensed and fragmented nuclei formation, a characteristic of apoptosis. Addition of 100 μM sodium sulfide caused almost complete protection against peroxynitrite-induced apoptosis, confirming the protective effects of ‘H2S’.
Hydrogen sulfide prevents apoptosis and protein nitration induced by peroxynitrite in HeLa cells
These results were also in agreement with the ability of hydrogen sulfide to prevent protein nitration induced by peroxynitrite. Namely, nitration of intracellular proteins in HeLa cells was induced either by 100 μM SIN-1 or 100 μM peroxynitrite and visualized by fluorescence microscopy. Addition of 100 μM ‘H2S’ almost completely blocked this effect (Figure 9B), confirming our in vitro results.
Finally, the ability of the product formed, HSNO2, to induce NO release inside of the cells was tested on HeLa cells using the fluorescent probe for NO, DAF-FM. Treatment with either 100 μM sodium sulfide, 100 μM decomposed peroxynitrite or 100 μM peroxynitrite caused no significant change in total fluorescence (see Supplementary Figure S7 at http://www.BiochemJ.org/bj/441/bj4410609add.htm). However, simultaneous addition of 100 μM sulfide and peroxynitrite significantly increased the fluorescence of the cells, an effect that was even stronger when HSNO2 (made by mixing 100 μM peroxynitrite and sodium sulfide) was added (Supplementary Figure S7), indicating that this product is indeed capable of releasing NO inside of the cells.
Previous studies have showed that hydrogen sulfide plays an important role as a signalling molecule in the regulation of blood pressure [1,2,4], but it also exhibits a strong protective role against ischaemia–reperfusion injury [11,12] and inflammation . However, the real biochemical mechanisms that underpin all these observations are still elusive as underlined in several reviews [1–3]. The role of peroxynitrite in inflammation and inflammation-related diseases, as well as in tissue damage in ischaemia–reperfusion injury, is well documented [15,32]. Thus in the present study we addressed both the chemical nature of the reaction of peroxynitrite with hydrogen sulfide and the physiological significance and consequence of such a reaction, in an attempt to shed light on possible mechanism(s) of the observed cytoprotective role of ‘H2S’.
The reaction between ‘H2S’ and peroxynitrite under aerobic and anaerobic conditions by both time-resolved UV–vis spectroscopy of peroxynitrite decomposition and by amperometry (following H2S consumption during the course of the reaction) was studied. The results of the present study clearly showed rapid consumption of peroxynitrite by hydrogen sulfide, which is different from results from a previous study , that used single-wavelength stopped-flow measurements. Time-resolved rapid-scan spectroscopy applied in the present study allowed us to observe the formation of a new product characterized by a visible absorption maximum at 408 nm. Furthermore, the rate constants for the peroxynitrite decomposition in the presence of ‘H2S’ were slightly higher than those observed by Carballal et al.  (4.8±1.4×103 M−1·s−1 at 37°C). However, the authors used NaHS as the source of the H2S/HS− equilibrium, which is highly hygroscopic and usually contains a high percentage of polysulfides ; another study  has also found a higher rate constant for the reaction of HOCl with H2S than reported by Caraballal et al. . On the basis of the rate constant and oxygen consumption, the authors also proposed the same reaction mechanism as previously observed for other thiols. However, the results shown in the present study speak in favour of a different mechanism.
Namely, the second-order rate constants for the reaction of peroxynitrite with cysteine, glutathione and the single thiol group of albumin at pH 7.4 and 37°C are 4.5×103 M−1·s−1, 1.35×103 M−1·s−1 [35,36] and 2.7×103 M−1·s−1  respectively. In comparison, the corresponding rate constant we obtained for the reaction of peroxynitrite with ‘H2S’ is somewhat higher (8±2×103 M−1·s−1 at 37°C). In the reaction of peroxynitrite with the thiol groups of cysteine and albumin, maximum oxidation yields were found at alkaline conditions, which led to the suggestion that the peroxynitrite anion, rather than peroxynitrous acid, was the main reactive species towards protonated thiols . In the reaction of low-molecular-mass thiols with peroxynitrite, the overall reaction is commonly presented as a two-electron oxidation by eqns (2 and 3):
If the same reaction scheme was applied to the reaction of ‘H2S’ with peroxynitrite, as proposed by Carballal et al. , the products of the reaction should be H2S2 and nitrite, but our UV–vis (Figures 1C and 1D and Supplementary Figure S1) and 15N-NMR (results not shown) spectra showed that nitrite is not initially formed (it appears with time when the product is exposed to oxygen, Supplementary Figure S2). H2S2, on the other hand, is a colourless compound with a particularly irritating odour that is unstable in aqueous solution and undergoes immediate decomposition followed by sulfur precipitation .
Peroxynitrite oxidation of thiols also occurs, at least partially, through one-electron-transfer oxidation processes, since the corresponding thiyl radicals were formed in peroxynitrite-mediated oxidation of cysteine, glutathione and the thiol group of albumin . In that mechanism, oxygen consumption has been observed (eqns 4 and 5):
However, our results clearly show that oxygen consumption is not involved in the initial reaction between H2S with peroxynitrite, which primarily results in a yellow product that decomposes subsequently upon oxygen consumption. Additional support for the observed reaction mechanism comes from the activation parameters obtained in the present study (Table 1), which indicate an associative interchange mechanism with formation of a product through simultaneous bond formation and bond breaking. Possible transformations involved in the rate-determining step could be presented as shown in 2.
Proposed transition state
Alternatively, one could consider the following reaction steps as well:
The reaction given in eqn (7) is highly thermodynamically favourable, with ΔG being more negative than −27 kcal/mol . At the same time, the kinetics of the reaction of sulfhydryl radical with oxygen given in eqn (8) and subsequently in eqn (9) is in a diffusion controlled range: 5×109 M−1·s−1 and 4×108 M−1·s−1 respectively . Therefore in such mechanism a dissociative process given in eqn (6) would be a possible rate-determining step. However, this dissociative mechanism would result in a significantly positive activation entropy and volume of activation, different from what we observed (Table 1). A possible partial involvement of these radical processes cannot be completely excluded, which could explain the lower yield of the product in the presence of oxygen.
Thionitrate, a reaction product with NO-donating properties
It has been documented that in the reaction of GSH with peroxynitrite, some NO is formed , possibly via formation of thionitrate, GSNO2 , and eventually thionitrite, GSNO . Formation of RSNO2 species has also been postulated to be a major link between nitroglycerine and its NO-donating properties .
Extensive computational studies performed by Thacher and co-workers [44,45] showed that three possible structural isomers of RSNO2 could exist: thionitrate (RSNO2), sulfenyl nitrite (RSONO) and sulfinyl nitrite [RS(O)NO], where R is CH3 or H. These structures are interchangeable, eventually leading to decomposition and NO release. However, experimental characterization of these isomers is still elusive. Our results offer for the first time the experimental support for such mechanism, as summarized in 1, which was originally proposed by Feelisch and Stamler  and Oae and Shinhama  more than a decade ago. Mass spectra indicate that the main product of the reaction is HSNO2 together with its decomposition product HSO−, and the observed NO release further supports the NO-donating properties of the initially formed HSNO2.
It should be mentioned that the DFT calculation by Artz et al.  predicted the UV–vis spectra of MeS(O)NO to display absorbance maxima at 287 and 418 nm, which corresponds well with the maxima observed for our reaction product (289 and 408 nm). Our calculation studies on HSNO2 isomers confirms that the only isomer with similar visible spectral properties is HS(O)NO. The initially formed sulfinyl nitrite could further react with a second molecule of ‘H2S’ and oxygen to give a species with higher oxidation states of sulfur like thiosulphate (S2O32−) (preliminary data, obtained by iodimetric titration) and nitrite, which could explain the faster decay of the product when exposed to air and the apparent stoichiometry of 2 moles of ‘H2S’ per 1 mole of peroxynitrite (eqn 10).
In the present study we also demonstrate that the product of the reaction possesses NO-donating properties, by the mechanism shown in 1. NO release seems to be higher at lower pH values (pH 6 and 5, Figure 4B), which are the pH values found in ischaemic tissue . Detection of NO in the cells simultaneously treated with peroxynitrite and ‘H2S’ or treated with HS(O)NO (Supplementary Figure S7) raises the question of a possible involvement of this process in the cytoprotection ascribed to ‘H2S’. It is worth mentioning that NO plays an important role in cardioprotection in ischaemia–reperfusion injury . It is proposed that NO-induced depolarization of the mitochondrial membrane potential protects cardiomyocytes by reducing the mitochondrial calcium overload during hypoxia–reoxygenation injury. Our results could offer an alternative route for generation of an intracellular NO donor. However, the real contribution of this pathway remains to be elucidated.
H2S does protect the cells from peroxynitrite-induced damage
Solely on the basis of the rate constant for the reaction of ‘H2S’ with peroxynitrite and taking into account even the pharmacological concentrations of ‘H2S’, it would be correct to conclude that ‘H2S’ would have a small, if any, physiological effect when compared with other thiols and/or CO2, as rightly proposed by Carballal et al. . However, our results demonstrate the strong protective effect of ‘H2S’ against peroxynitrite-induced damage under physiologically relevant conditions (such as presence of CO2) in both in vitro and cell culture conditions.
Both the oxidation and nitration of biomolecules are equally important reactions of peroxynitrite in the biological context, which is why we tested the ability of ‘H2S’ to modulate each of these processes. For example, MnSOD is a critical mitochondrial antioxidant enzyme: its nitration represents a severe hazard and has been suggested to promote oxidative damage and may ultimately signal to cell death . The complete in vitro blocking of peroxynitrite-induced MnSOD nitration (considered to be a main hallmark of ischaemia–reperfusion injury) observed in the present study already indicated that ‘H2S’ could indeed be capable of abrogating the detrimental effects of peroxynitrite. A definitive proof was provided in the cellular model where cells treated with ‘H2S’ showed diminished protein nitration when exposed to synthetic peroxynitrite or its donor, SIN-1 (Figure 9B).
As for the oxidation reaction, peroxynitrite is known to induce oxidation of a variety of biomolecules: proteins, lipids and DNA, with concomitant effects on cell metabolism and survival [15,16,32]. Peroxynitrite-induced DNA damage often results in cell death through either apoptois and/or necrosis, or it can be repaired by activation of PARP-1 . Following PARP-1 activation, high amounts of ADP-ribose are generated, which leads to the inactivation of the ABC transporter . Our results demonstrate that, even in the presence of CO2, the oxidation reaction is blocked in the presence of ‘H2S’, in a concentration-dependent manner. Furthermore, both apoptosis and necrosis induced by peroxynitrite were almost completely abolished by sulfide, whereas ABC transporter inactivation was partially prevented. This could explain the results of Elrod et al. , who have shown that ‘H2S’-caused cytoprotection against ischaemia–reperfusion injury of myocardium was associated with the inhibition of myocardial inflammation and with a preservation of mitochondrial structure and function, both states being directly dependent on peroxynitrite formation.
How can this discrepancy between in vitro kinetic data, which would favour the reaction of peroxynitrite with glutathione and CO2 (due to high intercellular concentration of both), and obvious strong protective effects of ‘H2S’ in vivo, be explained? Although the living cell is no longer referred to as a ‘bag of enzymes’ the extent to which the high internal concentration of macromolecules and the constraints of cellular architecture can influence intracellular biochemical reactions is still not generally appreciated . Since soluble macromolecules occupy a significant fraction of the total cell volume , intracellular space can be described as crowded or volume occupied . Hence, within such a crowded medium, the relative size and shape of a molecule and the probability of its successful diffusion, placement and effective contact with a potential target become crucial factors [52,53] that significantly alter its reaction rates . This speaks in favour of hydrogen sulfide when compared with GSH. Furthermore, Mathai et al.  showed that H2S diffuses through the cell membrane without a facilitator, i.e. a specific channel, making its diffusion even easier than that of water. In addition, gaseous H2S is shown to disappear very quickly from the plasma and its concentration is much lower in buffers that contain proteins , presumably due to electrostatic interactions of HS− ion with the positive charges of the protein (an introduction of a charge surrounding a molecule, for example, changes the reactivity towards ‘H2S’ from non-reactive to catalytically active; M.R. Filipovic and I. Ivanovic-Burmazovic, patent filed with the European Patent Office, 2011). All these above-mentioned phenomena could alternate the local concentration of ‘H2S’ and affect the kinetics of its reactions.
In conclusion, we show that peroxynitrite is rapidly consumed by ‘H2S’, and for the first time we shed light upon the nature of the initial reaction product between ‘H2S’ and peroxynitrite, which can be defined as HSNO2 and probably predominantly appears in its sulfinyl nitrite [HS(O)NO] form. Through this reaction, the pro-apoptotic, oxidative and nitrative properties of peroxynitrite are abolished. Furthermore, the thionitrate isomer formed is capable of releasing NO in a pH-dependent manner that could also partially account for the cytoprotective effects assigned to ‘H2S’ in ischaemia–reperfusion injury. Overall, these results suggest that a major part of the cytoprotective effects of hydrogen sulfide may be mediated by modulation of the peroxynitrite chemistry, in particular under inflammatory conditions. The unsolved preference of ‘H2S’ to react with peroxynitrite in vivo (where a variety of other thiols are present at much higher concentrations) highlights the pharmacological potential of ‘H2S’ donor synthesis, as their cytoprotective effects in a variety of inflammation-related diseases could be superior over the metal-based peroxynitrite decomposition catalysts.
carboxyfluorescein diacetate/succinimide ester
density functional theory
fetal bovine serum
manganese superoxide dismutase
poly(ADP-ribose) polymerase 1
Milos Filipovic and Ivana Ivanovic-Burmazovic conceived the study. Milos Filipovic, Jan Miljkovic, Andrea Allgäuer, Ricardo Chaurio and Martin Herrmann performed the experiments. Tatyana Shubina performed the computational studies and all authors analysed and discussed the data. Milos Filipovic and Ivana Ivanovic-Burmazovic wrote the paper with input from all the authors.
The authors would like to thank Professor Rudi van Eldik for his help with high pressure stopped-flow measurement and helpful discussions, Dr Leeane Nye for professional help with the high-resolution ESI–TOF MS, Jing Li for ESI-ion trap-MS measurements and Vladimir Prokopovic for technical assistance.
This work was funded by the Friedrich-Alexander-University Erlangen-Nuremberg and Deutsche Forschungsgemeinschaft [grant number SFB 583].
These authors contributed equally to this work.
Milos Filipovic and Ivana Ivanovic-Burmazovic have filed a patent with the European Patent Office for the change in reactivity of a protein towards H2S that occurs when a positive charge surrounding a molecule is introduced.