Chemical Biology of H₂S Signaling through Persulfidation

Abstract

Signaling by H₂S is proposed to occur via persulfidation, a posttranslational modification of cysteine residues (RSH) to persulfides (RSSH). Persulfidation provides a framework for understanding the physiological and pharmacological effects of H₂S. Due to the inherent instability of persulfides, their chemistry is understudied. In this review, we discuss the biologically relevant chemistry of H₂S and the enzymatic routes for its production and oxidation. We cover the chemical biology of persulfides and the chemical probes for detecting them. We conclude by discussing the roles ascribed to protein persulfidation in cell signaling pathways.

Graphical abstract

1. INTRODUCTION

Hydrogen sulfide (H₂S) is inextricably tied to the emergence of life on Earth. The recent discovery of unanalyzed samples from Miller’s 1958 experiment confirmed that sulfur-containing molecules (including the amino acids cysteine and methionine) could have been formed under the atmospheric conditions of early Earth from H₂S, which is released in volcanic emissions and from other geothermal activity.¹ It is postulated that RNA, protein, and lipid precursors have common origins in a cyanosulfidic protometabolism.² It is possible to create nucleic acid precursors on metal centers starting with hydrogen cyanide, H₂S, and ultraviolet light. Furthermore, the conditions that produced nucleic acid precursors likely also created the starting materials for natural amino acids and lipids suggesting that a simple set of reactions could have given rise to most of life’s building blocks.²

Early life forms likely thrived in an H₂S-rich environment. H₂S would have been useful for synthetic purposes but also as a source of metabolic energy. One of the first reported examples of lithothrophy, i.e., the ability to utilize inorganic substrates for energy generation, is of the H₂S-oxidizing bacterium Beggiatoa, discovered by Winogradsky.³ In this organism, H₂S provides the reducing power for CO₂ fixation via the Calvin cycle. Furthermore, green and purple sulfur bacteria use H₂S as an electron donor for photosynthetic CO₂ reduction. High levels of H₂S are lethal to most animals, but a few like pupfish, poeciliids, molluscs, and giant tubeworms are specialized to flourish in H₂S-rich habitats like marshes and deep-sea hydrothermal vents.

In medicine, perhaps the earliest reference to H₂S, even before its identity was established, was in a 1713 publication titled De Morbis Artificum Diatribes (Disease of Workers) by the Italian physician, Bernardino Ramazzini.⁴ He described an occupation hazard that manifested as a painful inflammation in the eye in workers who were chronically exposed to an unknown “acidic vapor” while cleaning privies and cesspits. Ramazzini also noted that this acidic vapor was responsible for coating silver and copper coins in the pockets of ill workers with a black substance (presumably silver sulfide and copper sulfide).⁴ While experimenting with pyrite ore (FeS₂) and mineral acid, the Swedish pharmacist Carl Wilhelm Scheele generated H₂S, which he described as “sulfur air” (Schwefelluft) in 1777.⁵

The historical reputation of H₂S as a poisonous gas endured until 1996, when Abe and Kimura first demonstrated that H₂S plays a role as an endogenous neuromodulator.⁶ Kimura’s group was also the first to report that H₂S acts as a smooth muscle relaxant,⁷ although the beneficial effects of H₂S on blood vessels had been known for a while. Thus, several Russian publications in the 1960s reported the beneficial effects of H₂S baths on coronary vasodilation and peripheral blood circulation after reconstructive operations on major arteries,^8–10 and the effect of H₂S on isolated rabbit aorta was reported by Kruszina and colleagues in 1985.¹¹ The vasodilatory property of endogenously generated H₂S was demonstrated in mice lacking γ-cystathionine (CSE),¹² as they developed profound age-related hypertension, with some parallels to another gas signaling molecule, nitric oxide (NO^•). These early observations propelled the current explosion of research on H₂S biology and signaling.

Another serendipitous discovery that put H₂S in the spotlight was the report that exposure of mice to subtoxic H₂S levels (20–80 ppm) decreased energy expenditure within a few minutes and induced a suspended animation-like state.¹³ The body temperature dropped by almost 20 °C, and the respiration rate decreased to 10% of normal. Remarkably, these effects were completely reversible, and the animals showed no apparent deficits upon recovery.¹³ This observation has spurred interest in the potential therapeutic development of H₂S to “buy time” for treating trauma patients.¹⁴

The past decade has witnessed a burgeoning literature on the physiological effects of H₂S and its role in many disease states, which are covered in several excellent reviews.^15–17 The proposal that signaling by H₂S involves postranslational modification of cysteine residues (i.e., Cys-SSH) provided a framework for understanding its physiological and pharmacological effects.^18,19 Protein persulfidation (erroneously described as sulfhydration) is also involved in biosynthetic pathways that require sulfur transfer, e.g., iron–sulfur clusters, biotin, thiamine, lipoic acid, molybdopterin, and sulfur-containing bases in RNA. The presence of the persulfide modification at a proteomic level was first examined only recently.^18–20

Due to its inherent instability, persulfide chemistry remains understudied. In this review, we introduce the biologically relevant chemistry of H₂S, cover the enzymatic routes for its production and oxidation, discuss the chemical biology of persulfides and review progress on the development of chemical probes for persulfide labeling and visualization. We conclude by discussing how persulfidation can control protein function and cell signaling pathways.

2. CHEMICAL PROPERTIES OF H₂S

H₂S is a flammable gas with the smell of rotten eggs. The water–H₂S system strictly obeys Henry’s law.^21–23 Some basic physicochemical properties of H₂S are given in Table 1. H₂S is a highly toxic gas. The human nose is considered to be one of the most sensitive H₂S sensors with a detection threshold of 0.02–0.03 ppm.^24,25 At 10 ppm, H₂S leads to eye soreness;²⁶ 20 ppm is the maximal allowable concentration for a daily 8 h exposure,²⁷ while exposure to 50 ppm of H₂S lead to conjunctival and mild respiratory irritation.^24,27,28 At 100 ppm, H₂S leads to olfactory loss within 3–15 min,²⁷ 150 ppm to olfactory nerve paralysis,^24,27,28 and exposure to 300–500 ppm represents an imminent threat to life leading to pulmonary edema.²⁹ Exposure to ≥500 ppm of H₂S leads to rapid loss of consciousness, cessation of respiration, and death.³⁰

Table 1.

Basic Physicochemical and Thermodynamic Properties of H₂S

dipole moment	0.97 D
boling temperature	−60 °C
solubility (in H2O)	110 mM/atm, 25 °C
	210 mM/atm, 0 °C
boiling temperature	−60.2 °C
density (25 °C, 1 atm)	1.36 kg/m3
IRa	ν1 2525, 2536 cm−1
	ν2 1169, 1184, 1189 cm−1
	ν3 2548 cm−1
1H NMRb	0.52 ppm
pK1	6.98
pK2	>17 at 25 °C
λmax (HS−)	230 nm
ε	8 × 103 M−1 cm−1
Henry’s law coefficient (298 K)	0.087135 mol solute/mol water atom
detection threshold by human nose	0.02–0.03 ppm
lethal dose	>500 ppm
ΔfG°(H2S)	−28 kJ/mol
ΔfG°(HS−)	+12 kJ/mol
ΔfG°(S2−)	+86 kJ/mol
E°′(S•−, H+/HS−)	+0.91 Vc
E°′(HS2−, H+/2HS−)	−0.23 Vc

^aValues are for the crystalline phase III.

^bValue obtained from crude sulfane oil.

^cVersus SHE.

2.1. Nomenclature

Formerly called hydrosulfuric acid or sulfhydric acid due to the acidic nature of its aqueous solutions, dihydrogen sulfide and sulfane are now the names recommended for H₂S by IUPAC. For HS⁻, the IUPAC recommended names are sulfanide or hydrogen(sulfide)(1−); for S²⁻, sulfide(2−), or sulfanediide. The term “H₂S” is used in this review for the gas and for the mixture of H₂S and HS⁻ in aqueous solution at a certain pH, unless otherwise specified.

The term “sulfanes”, according to the IUPAC Gold Book, includes polysulfanes, hydropolysulfides, and polysulfides, but its use is discouraged to avoid confusion with the newer systematic name sulfane for H₂S and the names derived therefrom. In the literature on the biological effects of H₂S, the term sulfane sulfur, sometimes abbreviated S⁰, is used to refer to a sulfur atom that is covalently bonded to two or more sulfur atoms (e.g., RS(S)_nSR, where (S)_n represents sulfane sulfurs) or to a sulfur atom and an ionizable hydrogen (e.g., Cys-SSH).^31,32 Some compounds containing sulfane sulfur are thiosulfate (S₂O₃²⁻ or ⁻S-SO₃⁻), persulfides (RSSH), inorganic and organic polysulfanes (HSS_nSH, RSS_nSR, and RSS_nH), polythionates (⁻SO₃–S_n–SO₃⁻), and cyclooctasulfur (S₈). Sulfane sulfur has six valence electrons in contrast to sulfide sulfur, which has eight, and is incorrectly referred to as “zero valence” sulfur, although it is always attached to other sulfur atoms or to an ionizable hydrogen. Sulfane sulfur can also be defined as sulfur that can tautomerize to the thiosulfoxide form (i.e., RSSH to RS(S)H). Sulfane sulfur usually has an oxidation state of zero.^33,34 It can be transferred to cyanide (CN⁻) to form thiocyanate (SCN⁻) and it can be reduced to H₂S by thiols (RSH). In this review, the term sulfane sulfur will be used to refer to a sulfur covalently bonded to two or more sulfur atoms or to a sulfur atom and an ionizable hydrogen. Elemental sulfur, that can be present in many different allotropic states of which the most abundant is S₈, will be abbreviated S_n.

2.2. Physicochemical Properties of H₂S

H₂S is a covalent hydride. Its structure is analogous to that of water, the hydride that is formed with oxygen, the companion to sulfur in the chalcogen group together with selenium and tellurium. However, the bond angles in H₂S are smaller than in water (93 versus 104°).³⁵ The frontier orbitals for the bent H₂S molecule are well described.^35,36 The molecular orbitals for H₂S result from the linear combination of the 1s orbital of the hydrogen atom and the 3s and 3p orbitals of the sulfur atom.^35,37 The energies of orbitals for H₂S versus HS⁻ are an important feature that defines the differences in their reactivity. For example the HOMO orbital of HS⁻ is less stable (−2.37 eV calculated and −2.31 eV measured)^38,39 than of H₂S (−10.47 eV) indicating that HS⁻ is more nucleophilic and basic than H₂S, which is consistent with their known reactivities. Because the HOMO is so stable, H₂S is not an excellent one-electron donor. The LUMO orbital for H₂S (+0.509 eV calculated and −1.1 eV based on electron affinity data) suggests that H₂S can be an excellent electron acceptor.^35,37 However, because LUMO is an antibonding orbital in the bent H₂S, electrons added to this orbital cause a weakening of both S–H bonds.³⁵

Interestingly, H₂S forms relatively strong hydrogen bonds with HS⁻ in aprotic solvents. At low temperatures H₂S reacts with triethylammonium hydrosulfide and with tetramethylammonium hydrosulfide to form complexes containing, respectively, 2 and 3 mol of H₂S per mole of salt. The energy of the hydrogen bond in HSH⋯SH⁻ is greater than 29 kJ/mol and possibly as large as 58 kJ/mol.⁴⁰ At pressures above 90 GPa (Gigapascal), H₂S becomes a metallic conductor of electricity. When subjected to extremely high pressures (∼1.5 million atmospheres (150 GPa)) and cooled below 203 K, H₂S displays the classic hallmarks of superconductivity: zero electrical resistance and a phenomenon known as the Meissner effect. The Meissner effect occurs when a superconducting material is placed in an external magnetic field and there is no field inside the sample, unlike in normal materials.⁴¹

H₂S has three low-temperature (ambient pressure) thermodynamic crystalline phases. IR spectra of crystalline phase III shows low-frequency bending vibration ν₂ at 1169, 1184, and 1189 cm⁻¹, higher frequency stretching vibrations, a symmetric stretch, ν₁ at 2525 and 2536 cm⁻¹, and an asymmetric stretch ν₃ at 2548 cm⁻¹.⁴²

Sulfur is larger than oxygen (covalent radius of 105 against 66), has a lower electronegativity (2.58 against 3.44 in the Pauling scale), and is more polarizable. As a consequence, the dipolar moment is lower for H₂S than for water (0.97 versus 1.85 D) and the intermolecular interactions are weaker. Thus, H₂S is a gas at room temperature and normal pressure, while water is a liquid (boiling points of −60 °C versus 100 °C). Nevertheless, H₂S has relatively high solubility in water (110 mM atm⁻¹ at room temperature and 210 mM atm⁻¹ at 0 °C).^43,44

H₂S is a weak diprotic acid (eqs 1 and 2).

$${\mathrm{H}}_2\mathrm{S}\rightleftharpoons {\text{HS}}^{-}+{\mathrm{H}}^{+}$$ (1)

$${\text{HS}}^{-}\rightleftharpoons {\mathrm{S}}^{2-}+{\mathrm{H}}^{+}$$ (2)

In aqueous solution, the pK_a valuesof the first dissociation (eq 1) are 6.98 and 6.76 at 25 and 37 °C, respectively. Different values for the second dissociation constant for HS⁻ have been reported. The original data indicated a pK₂ value at 25 °C ranging from 12.5 to 15.^43,45–49 However, Giggenbach pointed out that polysulfides formed at higher pH due to the oxidation of HS⁻ interfere with the determination of pK₂.^50,51 Based on optical spectra of highly alkaline, oxygen free, HS⁻ solution, Giggenbach estimated pK₂ to be 17 ± 0.1 at 24 °C.⁵¹ Meyer et al., confirmed this based on Raman spectroscopic monitoring of the H−S stretch in an oxygen-free HS⁻ solution with sodium hydroxide concentrations ranging from 5 to 22 M.⁵² Licht and Mansen proposed 17.3 for pK₂ of H₂S, based on pH measurements of highly alkaline K₂S solutions.⁵³ Using weak acid theory, which predicts a difference of 12.3 between pK₁ and pK₂ for an acid in which the negative charge resulting from the first dissociation step is localized on the same atom to which the second proton is bonded, Myers calculated a pK₂ value of 19 ± 2.⁵⁴ Extrapolating from the thermodynamic data for the dissociation of polysulfides and avoiding the experimental and theoretical difficulties associated with measurements in highly alkaline HS⁻ solutions, Schoonen and Barnes calculated pK₂ to be 18.51 ± 0.56.⁵⁵ Thus, although reference to the original values for pK₂ (12–15) persists in the biochemical literature, it is in fact higher (17–19).

At the physiological pH of 7.4 and at 37 °C, H₂S is in fast equilibrium with HS⁻, and the proportions of HS⁻ and H₂S are 81 and 19%, respectively. The concentration of S²⁻ is negligible (1.7 × 10⁻¹² M) but still sufficient to cause precipitation of metal sulfides, due to very low product solubility constants. Solutions of H₂S in water are mildly acidic with a pH of ∼4, solutions of NaHS are alkaline, and solutions of Na₂S are strongly alkaline. This information needs to be taken into consideration when adding H₂S or its salts to chemical or biological assays. The concentration of the HS⁻ anion can be determined from its absorption at 230 nm,⁵⁶ using a molar absorptivity of 8000 M⁻¹ cm⁻¹. However, air oxidation and polysulfide formation can complicate accurate determination.

2.3. Concentration in Membranes and Permeation of H₂S

The signaling actions of H₂S in compartments where it is not generated will be greatly influenced by its ability to concentrate in and diffuse across membranes. The partition coefficients (i.e., the ratio of its concentration in organic solvent/buffer) of H₂S between the organic solvents octanol or hexane and water are 2.1 ± 0.2 and 1.9 ± 0.5, respectively, at 25 °C and pH 3.8, a pH where the diprotonated H₂S form predominates.⁵⁷ These values indicate that H₂S is slightly hydrophobic since it is twice as soluble in organic solvents as in water. When the pH is increased to the more physiological value of 7.4, the partition coefficient decreases to 0.64 ± 0.05 (for octanol), due to the ionization of H₂S to HS⁻ in the aqueous phase.⁵⁷ Consistent with the values in organic solvents, the partition coefficient between dilauroylphosphatidylcholine liposomes and water is 2.0 ± 0.6 (pH 3.8, 25 °C).⁵⁷ This relatively high solubility in a membrane model is consistent with the high permeability of H₂S across biological membranes. Experimental estimates, comparison with other molecules, and molecular dynamics studies suggest that membrane permeability is as high as 11.9 cm s⁻¹ and that aquaporins or other protein facilitators are not needed for H₂S to cross membranes.^57–59 Nevertheless, according to mathematical models, biological membranes are expected to slow down H₂S transport resulting in local increases at sites of formation.⁵⁷

2.4. Reactivity of H₂S

The ability of HS⁻ to donate a pair of electrons and form a covalent bond, i.e., its nucleophilicity, is very good. This can be explained by its negative charge, by its high polarizability, and by the relatively low electronegativity of sulfur. Furthermore, HS⁻ is highly available at neutral pH due to the pK₁ value of H₂S being ∼7.

A generic reaction of HS⁻ with an electrophile (E₁⁺) is represented in eq 3. In contrast to the analogous reactions of thiolates (RS⁻, where the sulfur is bound to a carbon), the product formed from the reaction of HS⁻ with an electrophile can ionize and react with a second electrophile (E₂⁺) leading to a distinct product (eq 4). This differential reactivity between HS⁻ and thiolates is the basis of several methods for H₂S detection (see section 3).

$${\text{HS}}^{-}+{{\mathrm{E}}_1}^{+}\to {\mathrm{E}}_1-\text{SH}$$ (3)

$${\mathrm{E}}_1-{\mathrm{S}}^{-}+{{\mathrm{E}}_2}^{+}\to {\mathrm{E}}_1-\mathrm{S}-{\mathrm{E}}_2$$ (4)

The measure of nucleophilicity is a kinetic one and is estimated by comparing reaction rates, i.e. the faster the reaction, the greater the nucleophilicity. In this regard, it is interesting to compare the nucleophilicity of HS⁻ with that of alkyl thiolates (RS⁻). This comparison is biochemically relevant, since thiolates are abundant in biological systems. The rate constants of the reactions of HS⁻ with different disulfides (eq 5) are about 1 order of magnitude smaller than the corresponding reactions of thiolates (eq 6).⁶⁰

$${\text{HS}}^{-}+{\mathrm{R}}_1{\text{SSR}}_2\to {\mathrm{R}}_1\text{SSH}+{\mathrm{R}}_2{\mathrm{S}}^{-}$$ (5)

$${\text{RS}}^{-}+{\mathrm{R}}_1{\text{SSR}}_2\to {\mathrm{R}}_1\text{SSR}+{\mathrm{R}}_2{\mathrm{S}}^{-}$$ (6)

Equation 6 represents a thiol disulfide exchange reaction. These reactions occur through a concerted mechanism in which the attack of HS⁻ or RS⁻ on one of the sulfurs in the disulfide is accompanied by the release of the other as a thiolate. The lower rate constants in the case of HS⁻ versus RS⁻ can be attributed to the lack of an inductive effect by the adjacent methylene, to differences in polarizability, or to solvation effects.⁶⁰ Accordingly, computational calculations show that the energy of the highest occupied Kohn–Shan orbital, an indicator of nucleophilicity, is lower for HS⁻ than for thiolates, while the chemical hardness is higher.⁶⁰ The reactivity of HS⁻ toward hydrogen peroxide and peroxynitrite is also lower than that of thiolates.^61,62

The two-electron reduction potential E°′(HS₂⁻, H⁺/2HS⁻) is −0.23 V (versus SHE), which means that H₂S is a strong reductant.^63,64 The value is similar to the potentials for the cysteine and glutathione redox couples.^63,64 Importantly, the reaction of H₂S with two-electron oxidants such as hydroperoxides does not yield a disulfide (HSSH/HSS⁻) directly. Instead, sulfenic acid (HSOH) is formed as an intermediate (see section 5).

The one-electron reduction potential E°′(S^•−, H⁺/HS⁻) is estimated to be +0.91 V based on the thermodynamic parameters for these two species: Δ_fG°(S^•−) = +140 kJ/mol, Δ_fG°(HS⁻) = +12 kJ/mol, pK_a (HS^•) = 3.4⁶⁴ and is identical to the experimentally determined value of 0.92 V.⁶⁵ The value compares well with the values for thiols (E°′(RS^•, H⁺/RSH) = +0.96 V).^63,64 Given that pK₁ of H₂S is ∼7, the Gibbs energies of formation of H₂S and HS⁻ are identical at pH 7. The bond dissociation energy of H₂S is 90 kcal mol⁻¹ or 377 kJ mol⁻¹.⁶⁶ For comparison, the bond dissociation energy of H₂O is 118 kcal mol⁻¹ or 494 kJ mol⁻¹ and the one-electron reduction potential is E°′(HO^•, H⁺/H₂O) = +2.32 V.⁶⁷

The one-electron oxidation of H₂S to the sulfyil radical (HS^•) by biological oxidants is expected to be difficult given the high reduction potential of the S^•−/HS⁻ couple. Yet, H₂S is known to decay in air and is also oxidized by metals. The discrepancy is explained by the high reactivity of the resulting HS^•/S^•− radicals. HS^• (λ_max = 240 nm) dimerizes to give H₂S₂,^65,68 which in turn, readily decomposes to give S_n and H₂S (eq 7, 8),^69,70 both of which are removed from the system pulling the redox equilibrium in the direction of H₂S oxidation.

$${\mathrm{S}}^{•-}+{\text{HS}}^{•}\to {{\text{HS}}_2}^{-}k=9\times {10}^9{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$ (7)

$${{\text{HS}}_2}^{-}+{\mathrm{H}}^{+}\to \frac{1}{n}{\mathrm{S}}_{n(s)}+{\mathrm{H}}_2\mathrm{S}(\mathrm{g})$$ (8)

At pH > 5, ^•SH/S^•− reacts with HS⁻ (k_f = 4 × 10⁹ M⁻¹ s⁻¹, k_r = 5 × 10⁵ s⁻¹) to form disulfanuidyl (or dihydrogen disulfide radical anion), HSSH^•−/HSS^•2− (λ_max = 380 nm) (eq 9).^65,68

$${\mathrm{S}}^{•-}+{\text{HS}}^{-}\rightleftharpoons {\text{HSS}}^{•2-}{k}_f=4\times {10}^9{\mathrm{M}}^{-1}{\mathrm{s}}^{-1},k_r=5.3\times {10}^5{\mathrm{s}}^{-1}$$ (9)

Both HS^•/S^•− and HSSH^•−/HSS^•2− react with oxygen (eq 10–12). HSS^•2− is a weaker oxidant than S^•− (E°′(HSS^•2−, H⁺/HS⁻) = +0.67 V). EPR studies have identified the ^•SH radical in irradiated glassy solutions of sulfides and determined that its reaction with O₂ leads to formation of OSO^•− (λ_max = 255 nm) and not ⁻SOO^•.⁷¹

$${\mathrm{S}}^{•-}+{\mathrm{O}}_2\to {{\text{SO}}_2}^{•-}k=5\times {10}^9{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$ (10)

$${{\text{SO}}_2}^{•-}+{\mathrm{O}}_2\to {\text{SO}}_2+{{\mathrm{O}}_2}^{•-}k=1\times {10}^8{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$ (11)

$${\text{HSS}}^{•2-}+{\mathrm{O}}_2\to {\text{HSS}}^{-}+{{\mathrm{O}}_2}^{•-}k=4\times {10}^8{\mathrm{M}}^{-1}{\mathrm{s}}^{-1}$$ (12)

A major technical problem while working with H₂S solutions is the propensity for autoxidation. When H₂S or Na₂S are added to oxygen-free water a clear solution is formed. If the solution contains oxygen and trace metals in the pH range 6–9, a yellow-green color develops. The intensity of the color depends upon the concentration of elemental sulfur, S_n. Upon acidification, a whitish colloidal sulfur suspension forms.

Although the one-electron reduction of oxygen by HS⁻ (eq 13) is not thermodynamically favored, as reflected by the reduction potential E°′(O₂/O₂^•−) = −0.35 V (−0.18 V for a 1 M O₂ standard state), H₂S oxidation nonetheless occurs, albeit slowly.^72–76

$${\text{HS}}^{-}+{\mathrm{O}}_2\to {\mathrm{S}}^{•-}+{{\text{HO}}_2}^{•}\Delta E^{\circ }\prime =-1.26\mathrm{V}$$ (13)

The reaction between H₂S and O₂ is expected to be very slow from a kinetic point of view due to a spin barrier. O₂ has a triplet electronic ground state and a diradical character, which promotes its reactions with species with unpaired electrons but slows its reactions with species with paired electrons as in H₂S. The oxidation of H₂S (HS⁻ and S²⁻) by O₂ has a complicated stoichiometry with an array of products and metastable intermediates being produced. Products of all sulfur oxidation states have been reported: polysulfide ions (S₄²⁻ and S₅²⁻), sulfur (colloidal or orthotrombic), S₄O₆²⁻), S₂O₃²⁻, SO₃²⁻, S₂O₆²⁻and SO₄²⁻. Elemental sulfur, sulfite, sulfate, and thiosulfate are the major product observed in many studies and are usually formed in the stoichiometries shown in eqs 14–17:^72–76

$$8{\mathrm{H}}_2\mathrm{S}+4{\mathrm{O}}_2\to {\mathrm{S}}_8+8{\mathrm{H}}_2\mathrm{O}$$ (14)

$${\mathrm{H}}_2\mathrm{S}+\frac{3}{2}{\mathrm{O}}_2\to 2{\mathrm{H}}^{+}+{{\text{SO}}_3}^{2-}$$ (15)

$${{\text{SO}}_3}^{2-}+\frac{1}{8}{\mathrm{S}}_8\to {\mathrm{S}}_2{{\mathrm{O}}_3}^{2-}$$ (16)

$${\mathrm{H}}_2\mathrm{S}+2{\mathrm{O}}_2\to 2{\mathrm{H}}^{+}+{{\text{SO}}_4}^{2-}$$ (17)

Transition metal ions and complexes are effective catalysts as they are able to lower the activation energy for redox reactions.⁷⁷ This chemistry is exploited during industrial removal of H₂S, which is a corrosive gas, from sour waters and wastewaters, from gaseous streams, and from raw oil. For large industrial scale cleaning applications (especially for sour gases), the Claus process is employed, converting H₂S to S_n in two steps. In the first step, H₂S gets oxidized to SO₂, which symproportionates in a second reaction with another mole of H₂S to elemental sulfur of high purity (>99.5%).^78,79

For smaller applications and quantities, diverse setups and methods have been employed and patented. Some well-described applications include H₂S removal by bacteria, ultrasonic irradiation⁸⁰ bare iron or iron oxide surfaces,^81,82 iron solutions (Fe₂(SO₄)₃), or iron chelated agents (e.g., EDTA and CDTA).^83–85 Interestingly, addition of the heavy metal ion chelator DTPA (diethylenetriaminepentaacetic acid), which unlike EDTA completely chelates iron, stabilizes H₂S in solution and prevents its oxidation. The use of DTPA is highly recommended when working with H₂S (Na₂S or NaHS) solutions.⁸⁶

Inorganic polysulfides and sulfur formed during H₂S oxidation could have biological effects of their own. Inorganic polysulfides and their biological effects are covered in detail in section 8.4.

The chemistry of sulfur has been reviewed extensively.^87–89 Sulfur exists in many allotropic forms of which cyclic S₈ is the most stable.^87–89 In polar solvents such as methanol or acetonitrile, S₈ is partially transformed to S₇ and S₆ at ambient temperatures.⁹⁰ In aprotic solvents, hydroxide ion reacts with elemental sulfur S₈ to give the trisulfur anion radical S₃^•− as the major product.⁹¹ S₃^•− is highly reactive (see section 8.5). The allotropic composition of elemental sulfur is further perturbed by light.⁹² Sulfur can also exist as polymeric sulfur (S_∞). During oxidation of H₂S solutions, sulfur sol can arise, which consists of a sulfur core with hydrophilic polythionate (S_x(SO₃⁻3)₂) tails that enhance solubility and give rise to the characteristic yellow color.^87–89 Sulfur also has biological effects akin to H₂S. Intravenous injection of sulfur in rabbits led to the immediate detection of H₂S in breath.⁹³ Addition of colloidal sulfur to liver extracts led to its reduction to H₂S and to increased oxygen uptake.⁹⁴ Red blood cells can reduce sulfur in an NADPH- and glutathione-dependent manner, leading to H₂S release.⁹⁵ Some elemental sulfur preparations have entered preclinical trials recently, as H₂S donors.⁹⁶

3. WORKING WITH H₂S

Several methods are available for the qualitative and/or quantitative detection of H₂S. Before describing the different analytical methods, important considerations such as H₂S source, handling and safety precautions, and possible interferences in the measurements are discussed.

3.1. Handling Precautions

Although H₂S is relatively soluble in water and the pH dependent ionization to HS⁻ and S²⁻ increases the concentration of total H₂S species in the aqueous phase, solutions of H₂S or its salts, NaHS and Na₂S, lose H₂S to the gas phase. This loss is more significant when solutions are acidic rather than alkaline (pK_a of H₂S = 6.98, 25 °C) and when containers have large headspaces. Therefore, it is necessary to use sealed vials and to transfer H₂S-containing solutions using gastight syringes.⁹⁷ In addition, since H₂S tends to oxidize, particularly in the presence of metal ion contaminants, it is necessary to prepare solutions in anaerobic water or buffers, free of trace metals.⁹⁷ Storage of H₂S solutions is not recommended and solutions should be prepared immediately before use.

H₂S is highly toxic and should be handled in fume hoods. Investigators should not rely on their sense of smell for monitoring H₂S, because, although it can be detected at concentrations as low as 0.02 ppm, the inability to smell H₂S is one of the first signs of H₂S toxicity. Before discarding H₂S-containing solutions, a quenching solution containing zinc acetate (30 g/L), sodium citrate (9 g/L), and NaOH (12 g/L) can be used that results in insoluble ZnS formation. The safety aspects of working with H₂S have been reviewed recently.^86,98

3.2. Inorganic Sources of H₂S for Reference and Experiments

H₂S from commercially available gas cylinders can be a very pure source of this gas. Solutions are usually prepared by dissolving the gas in a deoxygenated solvent. A saturated solution containing ∼0.1 M H₂S and with a pH of ∼4 can be prepared in water. HS⁻ solutions can be prepared in buffer solutions with pH ∼9. The gas flow-through should be trapped as ZnS to avoid H₂S release into the atmosphere.⁸⁶ Alternatively, H₂S solutions can be prepared by mixing sulfide salts with acid in variations of Kipp’s apparatus for the preparation of gases.⁹⁹ In addition, concentrated solutions of H₂S (0.8 M) in tetrahydrofuran are commercially available.

Sulfide salts rather than H₂S gas are in fact often used for practical reasons. The salts that are usually used are sodium hydrogen sulfide (NaSH·xH₂O) and disodium sulfide, either anhydrous (Na₂S) or nonahydrate (Na₂S·9H₂O). The purity of the salts is an important consideration, particularly in the case of the NaSH salts.^{86,97,98,100,101} Being highly hygroscopic, these salts bind water from air and become liquid with time if kept outside of a glovebox. Of particular concern is the use of salts with unidentified numbers of water molecules (commercially available as NaSH·xH₂O, where x = 1–10), as it is unclear how researchers can calculate H₂S concentrations without a defined molecular weight. Recently the preparation of tetrabutylammonium hydrosulfide (NBu₄SH) was reported, which is a potentially useful source of HS⁻ in experiments performed in organic solvents.¹⁰² However, NBu₄SH is very hygroscopic and should also be handled with care in a glovebox. Frequent impurities found in all these sources are water, elemental sulfur, polysulfides, and other oxidation products such as sulfite and thiosulfate. Another concern is the alkaline pH of the solutions of the salts, particularly Na₂S, in water. The salts should be white, anhydrous powders and should be stored in desiccators under vacuum or in an argon box. It is convenient to wash the crystals with distilled water to remove oxidation products from the surface.⁸⁶ To eliminate sulfane sulfur contaminants, stock solutions can be treated with immobilized phosphines¹⁰³ or with cyanide.^100,101

Several natural and synthetic compounds can slowly liberate H₂S to potentially simulate its formation in biological systems. Various chemical groups and release mechanisms are involved, and this subject has been extensively reviewed.^104–110 NaSH and Na₂S salts are sometimes incorrectly referred to as H₂S donors; slow release of H₂S from them cannot be invoked. The acid base equilibration of these salts is extremely fast and consequently, the corresponding concentrations of HS⁻ and H₂S exist in solution with their ratio depending on the pH.

3.3. H₂S Donors

Recent advances in H₂S donor design have led to the development of several classes of donors that are showing very promising pharmacological effects. One main difference between these donors and sulfide salts is the slow release of H₂S, potentially mimicking physiological H₂S production. Significant problems with H₂S donors are that the chemistry of H₂S release is often unclear (for some, their pharmacologically similar effects to H₂S are used as an indication of their H₂S donor ability) and that the decomposition products could be reactive. The chemistry and biological applications of H₂S releasing agents has been extensively covered.^104–110 In this section, we summarize the main classes of H₂S donors, which are grouped based on the mechanism of their H₂S release: (i) donors that require thiols to release H₂S, (ii) donors that release H₂S by hydrolysis (with or without light), (iii) donors that release H₂S in reaction with bicarbonate, and (iv) COS-releasing donors that yield H₂S in the presence of carbonic anhydrase (Figure 1).

Figure 1.

Overview of different classes of H₂S donors. (A) Compounds based on the N-mercapto template (N-SH species) and the proposed mechanism for H₂S release. PG, protective group. (B) Perthiol-based compounds and proposed thiol-dependent mechanism of H₂S release. (C) Dithioperoxyanhydrides can also serve as H₂S donors upon reaction with thiols. (D) Arylthioamides release H₂S in the presence of thiols via an uncharacterized mechanism. (E) S-Aroylthiooximes release H₂S in the presence of aminothiols. (F) Chemical structures of Lawesson’s reagent and its derivative, GYY4137, the most widely used H₂S donor, and the proposed mechanism for H₂S release from GYY4137. (G) Phosphorothioate-based H₂S donors that release H₂S in a pH-dependent manner. (H) Another widely used class of molecules is 1,2-dithiole-3-thiones. They can be coupled to nonsteroid antiinflammatory drugs (NSAID) such as aspirin, ibuprofen, or naproxen, or to triphenylphsophonium group (AP39) which directs them to mitochondria. This class of molecules is believed to release H₂S via hydrolysis. (I) Example of photo cleavable gem-dithiol based H₂S donors, which undergo hydrolysis to release H₂S. (J) Thioamino acids release H₂S in reactions with bicarbonate. (K) COS, released by COS donors, forms H₂S in the presence of carbonic anhydrase. (L) Ammonium tetrathiomolibdate is shown to act as H₂S donor in vivo.

The simplest and oldest known thiol-activated H₂S donors are active principles of garlic.¹¹¹ Their chemistry is covered in section 8.6 as these compounds also release low molecular weight persulfides. The first synthetic thiol-activated donors were compounds based on the N-mercapto template (Figure 1A).¹¹² Since N-SH species are unstable, acyl groups were introduced to protect the mercapto group and enhance stability. In the presence of thiols such as GSH or cysteine, these compounds decompose to give H₂S. Similarly to N-SH donors, tertiary perthiol-based compounds were reported as H₂S donors, e.g., pencillamine-perthiol (Figure 1B).¹¹³ It is important to note, however, that H₂S release from these donors also results in the formation of mixed disulfides, which could introduce other modifications on proteins and initiate signaling. Nonetheless, the protective effects of pencillamine-perthiol based H₂S donors in myocardial ischemia/reperfusion injury have been reported. Dithioperoxyanhydrides were also recently reported as potential thiol-activated H₂S donors.¹¹⁴ Again, these compounds require 2 mol of thiol and release 1 mol of H₂S and a mixed disulfide (Figure 1C). Arylthioamides represent the fourth class of thiol-activated H₂S donors (Figure 1D). However, these compounds show very weak H₂S formation even in the presence of high concentrations of glutathione or cysteine.¹¹⁵ Nonetheless, when administered orally to rats, they induced a drop in blood pressure, reminiscent of the effect of H₂S. S-Aroylthiooximes have also been proposed as H₂S donors in the presence of aminothiols and show potential for the preparation of H₂S releasing materials (Figure 1E).^116,117

Some compounds, such as Lawesson’s reagent (2,4-bis(4-methoxyphenyl)-1,3,2,4-dithiadiphosphetane-2,4-disulfide) are known to release H₂S by spontaneous hydrolysis in aqueous solution.¹¹⁸ In fact, this compound was shown to be beneficial in regulating colon ulceration in a rat colitis model.¹¹⁹ A derivative of Lawesson’s reagent, GYY4137 (morpholin-4-ium metoxyphenyl(morpholino) phosphino-dithioate) is one of the most widely used H₂S donors (Figure 1F).^120–124 The pH-sensitive H₂S release (the lower the pH the greater the release) has been confirmed both colorimetrically and amperometrically, and the mechanism of H₂S generation recently elucidated.¹²⁵ This is a two-step process; the first, faster step involves straightforward sulfur–oxygen exchange with water, while the second, slower step, results in complete hydrolysis to an arylphosphonate (Figure 1F). GYY4137 has vasodilatory and anti-inflammatory effects akin to H₂S, and considering the time scale for most of its reported biological effects, the probable source of H₂S is the first reaction step (Figure 1F).¹²⁵ Using a core structure of GYY4137 or the phosphorothioate as a template, new compounds that undergo pH-dependent cyclization and subseqent H₂S release have been reported recently. Such donors could have particular application in ischemia-reperfusion injury where a pH drop is expected in ischemic tissues (Figure 1G).¹²⁶

1,2-Dithiole-3-thiones represent another class of H₂S-releasing molecules¹²⁷ that is widely used in the design of H₂S donors and is often coupled to some other pharmacologically active moiety, e.g., nonsteroidal anti-inflammatory drugs,^128–130 adenosine,¹³¹ or targeted to mitochondria with the lipophilic triphenylphosphonium cation (Figure 1H).^132–134 Hydrolysis is proposed to be involved in the mechanism of H₂S release, and the hydrolysis products were recently identified by mass spectrometry. Furthermore, this class of molecules was shown to directly persulfidate glutathione,¹³¹ while mitochondrially targeted AP39 (Figure 1H), even at nanomolar concentrations, increased intracellular persulfide levels more strongly than any H₂S donor.¹³⁵ Some NSAID conjugated hybrids have entered phase I clinical trial.¹³⁶ Some effort has been made in preparing photocleavable gemdithiol based H₂S donors, which then undergo hydrolysis to release H₂S (Figure 1I).¹³⁷

Thio-amino acids (thioglycine and thiovaline) reportedly react with bicarbonate and are converted to the corresponding N-carboxyanhydrides with concomitant release of H₂S (Figure 1J).¹³⁸ Considering the high concentration of bicarbonate in the biological milieu and its common use as a buffer in cell culture media, these compounds could potentially be useful as H₂S donors. Unlike other reported classes, these donors reach their plateau after 1 h and could be classified as donors with moderate-to-fast H₂S release.

Some compounds are able to release carbonyl sulfide (COS; e.g., thiocarbamates and N-thiocarboxyanhydrides). COS is converted into H₂S by the action of the ubiquitous enzyme carbonic anhydrase (Figure 1K).^139,140 This chemistry has been explored to create molecules that release COS (and subsequently H₂S) upon light activation or intracellular reaction with reactive oxygen and nitrogen species.^141,142

The first metal complex-based H₂S donor has been reported recently, ammonium tetrathiomolybdate (Figure 1L).¹⁴³ (NH₄)₂MoS₄ was shown to slowly release H₂S in buffers and cell culture¹⁴³ and exhibited cytoprotective effects in a rat model of ischemia-reperfusion injury.¹⁴⁴

3.4. Methods for H₂S Measurement

Unlike H₂S, the deprotonated species, HS⁻ and S²⁻, absorb in the ultraviolet region with absorption coefficients at 230 nm of 8.0 × 10³ and 4.6 × 10³ M⁻¹ cm⁻¹, respectively, at 25 °C.¹⁴⁵ In principle, H₂S solutions can be diluted in buffer at pH ∼9.6 (e.g., carbonate buffer), where H₂S is predominantly in the HS⁻ form, and the concentration can be determined from the absorbance at 230 nm.⁸⁶ This approximation is, however, useful only for very pure solutions since the presence of H₂S oxidation products as well as other components in the case of complex mixtures can cause interference.

To standardize stock solutions, classical iodometric titrations can be performed. For this, H₂S is trapped in zinc acetate to minimize its diffusion and then reacted with excess iodine in acidic medium. The remaining iodine is titrated with sodium thiosulfate, using starch as an indicator (eqs 18 and 19). However, the presence of other reductants can lead to error.

$${\mathrm{S}}^{2-}+{\mathrm{I}}_2\to \mathrm{S}+2{\mathrm{I}}^{-}$$ (18)

$${\mathrm{I}}_2+2{\mathrm{S}}_2{{\mathrm{O}}_3}^{2-}\to 2{\mathrm{I}}^{-}+{\mathrm{S}}_4{{\mathrm{O}}_6}^{2-}$$ (19)

The development of H₂S detection tools has been rapidly expanding and has been covered in several review articles. In the following sections, an overview of the most widely used methods for H₂S detection is provided.

3.4.1. Methylene Blue Method

This method is based on the synthesis of methylene blue from H₂S and N,N-dimethyl-p-phenylenediamine in the presence of acid and ferric chloride (Chart 1). The oxidative coupling of two molecules of the diamine with H₂S involves the initial one- or two-electron oxidation of the diamine to the cation radical or diimine intermediates, respectively, followed by nucleophilic attack of H₂S to form a thiophenol derivative that reacts with a second molecule of the oxidized intermediate.^146,147 Zinc chloride is also included in the assays to prevent H₂S volatilization. The concentration of methylene blue is measured at 670 nm and compared to calibration curves obtained with samples of known concentrations of H₂S that were similarly processed.^148–150 Variations of this method include chromatographic separation of methylene blue¹⁵¹ and mass spectrometric detection.¹⁵² The methodological details and potential pitfalls of this method have been reviewed.^32,101,153 Some of the key drawbacks are its low sensitivity (in the μM range), the release of sulfides from acid-labile stores (like iron sulfur clusters) that can lead to a gross overestimation of H₂S, limited linear range for absorbance of methylene blue on concentration, and the potential for interference due to the presence of thiols or due to the turbidity of biological samples.

Chart 1.

Measurement of H₂S through Formation of Methylene Blue

3.4.2. Lead Acetate

A simple way to follow the enzymatic synthesis of H₂S is to use lead acetate and measure the formation of insoluble lead sulfide by the increase in turbidity at 390 nm. H₂S concentrations are determined by comparison to a calibration curve generated with known concentrations of lead sulfide. Lead acetate is also useful for activity staining in gels; enzymes that produce H₂S are identified as dark spots by soaking native gels in solutions containing the appropriate H₂S-generating substrate(s) plus lead acetate.¹⁵⁴ An alternative approach is to use lead acetate-soaked filter paper and to measure the appearance of black spots using densitometry.^155,156 However, this method provides only semiquantitive data, and the sensitivity is quite low.

3.4.3. Electrochemical Sensors

Two types of sensors have been used: ion-selective electrodes and polarographic sensors. The H₂S ion-selective electrodes use a silver sulfide membrane that specifically interacts with S²⁻ creating a change in potential across the membrane. The electrodes are inexpensive, easy to operate, and highly selective. However, they require a long equilibration time and frequent reconditioning to remove interfering materials. Furthermore, they require a high pH that could interfere by releasing H₂S from proteins. Second, the polarographic H₂S sensors are based on measurement of the current produced from the oxidation of H₂S by ferricyanide.¹⁵⁷ These sensors contain a membrane through which H₂S permeates into an internal solution of alkaline potassium ferricyanide, where chemical oxidation of H₂S occurs concomitant with the reduction of ferricyanide to ferrocyanide. The latter is then reoxidized electrochemically. The polarographic sensors have a shorter response time and higher sensitivity, allowing for real-time monitoring of H₂S down to ∼10 nM. However, they have the tendency to leak easily and have large residual currents due to impurities. Enzyme-based electrochemical H₂S sensors have also been proposed and reviewed recently.¹⁵⁸

3.4.4. Gas Chromatography

The determination of H₂S in aqueous samples can be carried out following derivatization (e.g., to bis(pentafluorobenzyl)sulfide), extraction into organic solvents, and gas chromatography analysis with different detection systems.^32,101 Alternatively, gas chromatography can be used for the direct determination of H₂S gas removed from the headspace of reaction vessels and analyzed using a flame photometric detector or a sulfur chemiluminiscence detector that has high sensitivity.^32,159,160 The concentration of H₂S in the aqueous phase of the assay mixture at a given pH is then calculated based on mass conservation considerations knowing the pH of the solution, pK₁ for H₂S, and its solubility at the assay temperature. A big advantage of this method is that, when coupled to a sulfur chemiluminiscence detector, very low amounts of H₂S (0.5 pmol) can be measured. Handling of the samples, however, requires particular care and gastight equipment.

3.4.5. Monobromobimane Derivatization

The fluorogenic reagent monobromobimane, which was originally introduced to label thiols (RSH),¹⁶¹ can also be used to measure H₂S in the namolar range.^162–164 Both thiols and H₂S react with monobromobimane via a nucleophilic substitution process. In the case of thiols, a thioether is formed, while in the case of H₂S, a bimane-substituted thiol is formed, which can react with a second monobromobimane forming dibimane sulfide (Chart 2). The latter can be detected by its fluorescence during HPLC separation or by mass spectrometry. This method has found a broad use lately; however, an important consideration with monobromobimane-based detection is the relatively slow reaction rate (k ≈ 10 M⁻¹ s⁻¹ at pH 8).¹⁶² Monobromobimane has also been used to quantify polysulfides and persulfides by mass spectrometry.¹⁶⁵ The preparation of the corresponding standards is, however, challenging due to their instability; bimane polysulfides may be unstable too.

Chart 2.

Reaction of Monobromobimane with H₂S to form Dibimane Sulfide

3.4.6. Other Fluorescent Probes

The growing interest in detecting H₂S in biological samples is spurring the development of H₂S probes for tissue, cellular, and subcellular imaging. Such probes usually contain a fluorescence quencher that can be modified or removed by H₂S. Various reactions and molecular scaffolds have been used, with different reaction rates, selectivity, and potential limitations. Since these probes have been recently reviewed,^98,166–168 only selected examples are provided in Charts 3–5. One strategy is to detect the reduction of azide groups by H₂S to amines in a variety of scaffolds including rhodamine, dansyl, and naphthalimide scaffolds (Chart 3A).^169–171 Alternatively, the reduction of nitro groups to amines has been used¹⁷¹ (Chart 3B). A limitation of these probes is their slow reaction rates and the possible interference of other species, particularly thiols. Furthermore, the quantification of H₂S from biological samples is not really possible. Considering that most of the probes react with H₂S irreversibly, they actually remove sulfide from the intracellular pool and reach saturation rapidly giving a potentially incorrect impression of high intracellular steady state levels of H₂S. Future development of reversible probes would permit measurements of actual changes in H₂S levels.

Chart 3. Examples of Fluorescent Probes for H₂S Detection Based on the Reduction of Azide or Nitro Groups^a.

^a(A) Reduction of azide groups in rhodamine (top), dansyl (middle), and naphthalimide (bottom) scaffolds. (B) Reduction of nitro group in the naphthalimide scaffold.

Chart 4. Examples of Fluorescent Probes for H₂S Detection Containing Two Electrophilic Centers^a.

^a(A) Probe containing an aldehyde and an acrylate group on a triaryl pyrazoline scaffold. (B) Probe containing an activated disulfide on a fluorescein scaffold.

To improve the selectivity over thiols, the double nucleophilicity of H₂S can be exploited, as in the case of the reaction with monobromobimane (Chart 2). Probes have been developed that contain two electrophilic centers; for example, an aldehyde and an acrylate. H₂S first reacts with the aldehyde forming a thiohemiacetal group that then undergoes a Michael addition reaction with the acrylate group (Chart 4A).¹⁷² Alternatively, the two electrophilic centers can be a disulfide and an ester. In the example shown in Chart 4B, H₂S reacts with an activated disulfide forming a persulfide and a thiol. The persulfide then attacks the ester liberating a fluorophore and benzodithiolone.¹⁷³

Another property that can be exploited for H₂S detection is its high affinity for metals. Probes have been developed that contain a fluorophore bound to Cu²⁺, which quenches fluorescence. Binding of H₂S to the metal ion results in CuS precipitation and an increase in fluorescence (Chart 5).¹⁷⁴

Chart 5.

Example of a Probe for H₂S Detection Based on the Release of Copper Sulfide from a Cyclen and Fluorescein Derivative

3.5. Endogenous Concentration of H₂S

The methods described above can be used with appropriate precautions to detect and quantify H₂S in simple solutions. However, their use with biological samples can be confounded by side reactivity, leading to highly variable estimates of H₂S concentration. These differences can arise from the nature of the standards used, from sample loss due to the volatility and oxidation lability of H₂S, from interference from species with similar reactivity (e.g., thiols and persulfides), and from the release of H₂S from labile pools during sample handling (e.g., acidification, alkalinization, reduction). Biological samples contain labile sulfur compounds that release H₂S upon certain chemical treatments.^32,101 For example, exposure to acidic pH, which is associated with some analytical methods, liberates H₂S from iron sulfur clusters. Addition of reductants such as dithiothreitol liberates H₂S from sulfane sulfur compounds, particularly from persulfides, polysulfides, and elemental sulfur. Alkaline conditions result in H₂S release from various sulfur-containing species, particularly thiols and disulfides. This potential for introducing artifacts has contributed to the estimates of H₂S concentration in biological samples varying over 5 orders of magnitude.

As pointed out by Olson,^153,175 it is important to critically evaluate the reliability of the reported values for H₂S in biological samples. The concentration of H₂S in tissues is often expressed as moles of H₂S per gram protein or as moles of H₂S per gram of wet weight. Cells are typically 10–20% protein and 60–75% water. Thus, 1 nmol of H₂S (mg protein)⁻¹ is equivalent to ∼200 μM H₂S, and 1 nmol of H₂S (mg wet weight)⁻¹ is equivalent to ∼1500 μM. In cultured cells, H₂S concentration is sometimes expressed as moles per cell. Considering that a human cell has a volume of the order of 10⁻¹² L (i.e., 10³ fL or 10³ μm³), then 1 nmol of H₂S (10⁶ cells)⁻¹ is equivalent to ∼1000 μM. Since, the human nose can detect ∼1 μM H₂S in solutions,^159,176 many reports of H₂S concentrations in biological samples are undoubtedly in error.

Before 1996, when H₂S was recognized as a physiological mediator, essentially all measurements of blood H₂S had either failed to detect it or yielded extremely low values, consistent with the fact that H₂S cannot be smelled in blood. Since 1996, reports of blood H₂S concentrations had risen to an average of ∼50 μM.^175,176 However, the use of more sensitive methods, e.g., polarographic sensor or gas chromatography coupled to a sulfur chemiluminiscence detector, together with greater rigor in sample preparation, are revealing that the concentration of H₂S in blood is <100 nM and may be as low as ∼100 pM.^159,177

Gas chromatography coupled to chemiluminiscence detection has revealed that the basal tissue H₂S levels are quite low. According to one study, the basal H₂S level is ∼10−15 nM in murine liver and brain.¹⁵⁹ Another study reported levels of 0.004–0.055 μmol H_sS kg⁻¹ or 0.03–0.39 μmol (kg protein)⁻¹, corresponding to 6–80 nM, in murine liver, brain, heart, muscle, esophagus, and kidney.¹⁷⁸ In agreement with these low estimates, the steady-state concentration extrapolated from measurements of H₂S production and consumption rates in murine liver, kidney, and brain were calculated to be 9–29 nM.¹⁷⁹ Curiously, H₂S levels in aorta are significantly higher (∼1.5 μM).¹⁷⁸

The steady state concentration of H₂S is the net result of its formation and decay rates. Production rates have been estimated to be 0.45, 0.3, and 0.1 pmol min⁻¹ (mg tissue)⁻¹ (i.e., ∼0.6, 0.4, and 0.2 μM min⁻¹) in intact colon muscle, brain, and liver of mice in the presence of 10 mM cysteine; the production rate increased to 8, 7, and 20 pmol min⁻¹ (mg tissue)⁻¹ (i.e., ∼11, 10, and 28 μM min⁻¹) respectively, in homogenized tissue.¹⁸⁰ Another study reported H₂S production rates by murine liver and brain homogenates of 106 and 1.2 nmol min⁻¹ (g tissue)⁻¹ (i.e., ∼151 and 1.7 μM min⁻¹), respectively, in the presence of 10 mM cysteine.¹⁵⁹ At a more physiologically relevant concentration of 0.1 mM cysteine, H₂S production rates of 48 μmol h⁻¹ (kg tissue)⁻¹ (i.e., ∼ 12 μM min⁻¹) and 29 μmol h⁻¹ (kg tissue)⁻¹ (i.e., 0.7 μM min⁻¹) by murine liver and brain homogenates, respectively, were reported.¹⁷⁹ The decay rates of H₂S are high, and as expected, they decrease dramatically under hypoxic conditions.¹⁷⁹ The apparent first order rate constant of H₂S decay in murine liver under aerobic conditions was reported to be 277 min⁻¹ at 37 °C.¹⁷⁹ Thus, the very low steady-state tissue concentrations of H₂S are primarily due to the high rate of its oxidation.¹⁷⁹

4. ENZYMES INVOLVED IN H₂S BIOGENESIS

At least three enzymes in the mammalian sulfur metabolic network have the potential to synthesize H₂S.^181–185 Two of these enzymes, cystathionine β-synthase (CBS) and CSE, comprise the transsulfuration pathway. The latter provides an avenue for synthesizing cysteine from the essential amino acid methionine, via the metabolic intermediate, homocysteine. The transsulfuration pathway enzymes exist predominantly in the cytoplasm although, under some conditions, they are reportedly located in other compartments such as the nucleus¹⁸⁶ or the mitochondrion.^187,188 The third enzyme, mercaptopyruvate sulfurtransferase (MST), resides in both mitochondrial and cytoplasmic compartments.¹⁸⁹ It converts 3-mercaptopyruvate, derived from cysteine transamination, to pyruvate and transfers sulfur to a thiophilic acceptor forming a persulfide from which H₂S can be released. In this section, an overview of the recent literature on the structure, mechanism, and regulation of these three enzymes is discussed with an emphasis on the recent literature pertaining to H₂S biogenesis by the human enzymes.

4.1. Reactions Catalyzed by CBS

Located at the crossroads of the methionine cycle and the transsulfuration pathway, CBS commits sulfur to cysteine synthesis and catabolism, which in turn influences H₂S biogenesis. CBS exhibits substrate promiscuity and catalyzes a multitude of reactions at the β-carbon of the substrates, serine and cysteine (Chart 6).^{154,185,190–192} In its role in the transsulfuration pathway, CBS catalyzes the β-replacement of serine by homocysteine forming cystathioninine and eliminating H₂O (Chart 6, [1]). It can also catalyze the β-replacement of cysteine by homocysteine [2], by a second mole of cysteine [3], or by water [4], forming cystathionine, lanthionine, or serine, respectively, and eliminating H₂S in the process. Finally, CBS can catalyze the β-elimination of cystine forming cysteine persulfide (Cys-SSH) [5]. Mutations in CBS are the most common cause of hereditary homocystinuria, an autosomal recessive disorder.¹⁹³

Chart 6. Reactions Catalyzed by CBS^a.

^aReaction 1 generates cystathionine in the canonical transsulfuration pathway. Reactions 2–4 generate H₂S from cysteine and/or homocysteine, and reaction 5 produces Cys-SSH from cystine.

4.1.1. Organization of CBS and Properties of the Heme Cofactor

Human CBS is a homodimer with a subunit molecular weight of ∼63 kDa. Its propensity for aggregation leads to its isolation as higher order oligomers ranging from 4-to 16-mers. The crystal structures of full-length human^194,195 and Drosophila¹⁹⁶ CBS have been obtained for the dimers. CBS is unique in being the only known PLP enzyme that is also a hemeprotein.¹⁹⁷ It is a modular protein with an N-terminal domain spanning ∼70 residues, which binds a regulatory heme b cofactor (Figure 2A). This is followed by a middle catalytic core (spanning residues 71–411, human numbering), which houses the PLP cofactor and resembles the fold II or β-class of PLP enzymes.¹⁹⁸ The catalytic core is conserved across organisms regardless of whether CBS contains or lacks the heme domain. A Cys272-X-X-Cys275 motif present in the catalytic core is seen in the reduced dithiol and oxidized disulfide state in two structures^199,200 and could potentially render CBS sensitive to regulation by metal ions or to oxidation. CBS is reportedly inhibited by free copper (10–25 μM), although a connection between this observation and chelation by the CXXC motif has not been made.²⁰¹ The C-terminal domain (spanning residues 412–551) comprises a tandem repeat of two “CBS domains”, which is a β–α–β–β–α secondary structure motif found in diverse proteins that often binds adenosine derivatives and is associated with energy sensing.²⁰² In CBS, the C-terminal domain binds S-adenosylmethionine (AdoMet),²⁰³ an allosteric activator.²⁰⁴ Hence, ∼40% of the protein is involved in the N- and C-terminal regulatory domains, which exert allosteric control over the CBS-catalyzed reaction.

Figure 2.

Organization and structure of human CBS. (A) CBS is a modular protein with regulatory domains at its N- and C-termini. The C-terminal domain comprises a tandem repeat of two CBS domains, CBS1 and CBS2. The structures of human CBS in the absence (PDB: 4L27) (B) and presence (PDB: 4PCU) (C) of AdoMet show that a large conformational rearrangement accompanies the transition from the basal to the activated state. The protomers are shown in blue and yellow, respectively, the heme (red) and PLP (yellow) are in sphere representation, and the blue arrows point to the intervening linker region between the catalytic core and the C-terminal domain.

Three structures of full-length CBS have been reported: two of human CBS in the presence and absence of AdoMet^194,195 and a third of Drosophila CBS,¹⁹⁶ which does not bind AdoMet and exists in a hyperactivated state.²⁰⁵ These structures reveal that AdoMet binding elicits a remarkable conformational rearrangement. In the absence of AdoMet, an intersubunit crossover of the C-terminal domains places each by the active site entrance of the other subunit, impeding substrate access (Figure 2B). In the presence of AdoMet, the C-terminal domain dimerizes atop the catalytic domains (Figure 2C). This structural rearrangement explains why AdoMet binding²⁰⁴ or truncation of the C-terminal domain entirely,²⁰⁶ activates CBS, i.e., by facilitating substrate access to the active site.

Although held by an unstructured N-terminal loop and relatively exposed, the heme in CBS is tightly bound. The first structures of truncated CBS lacking the C-terminal regulatory domain revealed that Cys52 and His65 coordinate the heme iron (Figure 3).^199,200 The low-spin heme iron retains these ligands in both the ferric and ferrous oxidation states. Upon reduction, the Soret peak shifts from 428 to 448 nm while the α/β absorption bands shift from a broad feature centered at ∼550 nm (in ferric CBS) to 571 and 540 nm (in ferrous CBS).²⁰⁷ Early NMR, pulsed EPR,²⁰⁸ and resonance Raman^209,210 studies had ruled out a catalytic role for the heme. ³¹P NMR studies demonstrated that the spin–lattice relaxation rates in the paramagnetic ferric (6.34 ± 0.01 s) and diamagnetic ferrous (5.04 ± 0.06 s) states were similar, indicating that the PLP and heme cofactors are not proximal to each other.²⁰⁸ The crystal structures revealed that an ∼20 Å distance separates the heme and PLP sites (Figure 3)^199,200 and that this distance is not modulated by the presence or absence of the C-terminal regulatory domain or its allosteric effector, AdoMet.^194,195 Yeast CBS lacks the heme cofactor but is highly active,²¹¹ further arguing against a catalytic role for this cofactor. In fact, deletion of the N-terminal 69 residues in human CBS results in a heme-less variant, which albeit less stable, retains ∼40% of wild-type activity.²¹²

Figure 3.

Close up of the CBS structure. The interactions between the Cys52 heme ligand and Arg266 at one end of the α-helix and between Thr257 and Thr260 and the phosphate group of PLP at the other are shown. Asn149 hydrogen bonds with the C4 oxygen in PLP.

A rhombic EPR signal is associated with ferric CBS with g values of 2.5, 2.3, and 1.86, which is similar to that of model heme complexes and heme proteins with imidazole/thiolate ligands.²¹³ Resonance Raman studies using ³⁴S-labeled CBS identified the ν(Fe–S) vibration at 312 cm⁻¹.²⁰⁹ Exposure to mercuric chloride, a thiol chelator, resulted in CBS converting from a 6-coordinate low-spin to 5-coordinate high-spin state with a Soret maximum at 395 nm and a rhombic g = 6 EPR signal.²¹³ The spin-state change induced by mercuric chloride was correlated with a loss of CBS activity,²¹⁰ consistent with long-range communication between the heme and PLP sites. ³¹P NMR studies also provided evidence for long-range signal transmission by revealing that the chemical shift of the phosphorus nucleus in PLP shifted from 5.4 to 2.2 ppm upon reduction of the heme iron.²⁰⁸

The ferric heme in CBS, which is coordinately saturated, is relatively inert to exchange by exogenous ligands.²¹⁴ The reduction potential of the Fe³⁺/Fe²⁺ couple is −350 ± 4 mV for full-length CBS²¹⁵ and −291 ± 5 mV for truncated CBS²¹⁶ lacking the C-terminal regulatory domain. Following reduction, ferrous CBS can bind exogenous ligands such as CO, NO^•, cyanide, and various isonitriles.^217–219 The heme cofactor reduces nitrite to ferrous heme-bound NO^•.²²⁰ Binding of these ligands is associated with loss of activity, where characterized. Despite the low reduction potential for the heme iron in full-length CBS, it can be reduced by an NADPH-driven flavin oxidoreductase when coupled to carbonylation by CO, establishing the potential physiological relevance of this reaction.²²¹ Ferrous CBS does not bind O₂; instead it undergoes rapid oxidation (1.1 × 10⁵ M⁻¹ s⁻¹) apparently by an outer sphere mechanism generating superoxide radical and ferric heme.²¹⁶ Reoxidation of the ferrous-nitrosyl heme on CBS leads to peroxynitrite formation.²²² Thus, the heme redox activity makes CBS a potential source of both reactive oxygen (O₂^•−) and nitrogen (ONOO⁻) species.

AdoMet increases the affinity of the CBS heme for NO^• (2-fold) and CO (5-fold) and thereby sensitizes the enzyme to inhibition.²²³ Hence, in the interplay between the N- and C-terminal regulatory domains, heme supersedes AdoMet as an allosteric regulator. AdoMet activates ferric CBS but potentiates the inhibition of ferrous CBS by CO or NO^•. Together with the effect of the C-terminal domain on the heme redox potential discussed above, the effect of AdoMet on the affinity of the heme ligands, CO and NO^•, hints at long-range communication between the N- and C-terminal domains, which are >50 Å apart in the structure of AdoMet-bound CBS.¹⁹⁵

4.1.2. Catalytic Mechanism of CBS

The ping pong reaction cycle of CBS involves the following steps: (i) binding of the first substrate (serine or cysteine), which results in displacement of Lys119 and formation of the corresponding external aldimine, (ii) abstraction of the α-proton by Lys119 leading to a resonance stabilized carbanion, (iii) elimination of water or H₂S leading to aminoacrylate formation, (iv) addition of the second substrate (homocysteine, cysteine, or water) to give the corresponding product external aldimine, and (v) reformation of the Schiff base with Lys119 leading to product release (Chart 7). Support for this reaction mechanism has been obtained by stopped-flow kinetic studies on human²²⁴ and yeast²²⁵ enzymes as well as from kinetic studies on a heme-less variant of human CBS.²¹² Since the absorbance of the heme cofactor obscures the PLP cofactor, difference stopped flow spectrometry had to be used to monitor PLP-bound reaction intermediates in the human enzyme.²²⁴ Mutation of Lys119 to alanine reduces CBS activity ∼1000-fold and the exogenous base, ethylamine, leads to a 2-fold higher activity, consistent with the role of Lys119 as a general base in addition to its involvement in Schiff base formation.²¹²

Chart 7. Reaction Mechanism of CBS and Structures of Key Intermediates^a.

^a(A) A minimal mechanism is shown for the β-replacement of serine by homocysteine to generate cystathionine and water. Structures of the carbanion (B) and aminoacrylate (C) intermediates trapped in Drosophila CBS (PDB: 2PC4 and 2PC3) are shown. The numbering of residues shown in parentheses in B and C are for human CBS. The corresponding residues in the fly protein are Lys88, Ser116, Asn118, and Ser318, respectively. An sp² hybridized α carbon and sp² hybridized α and β carbons are seen in the carbanion and aminoacrylate intermediates, respectively. Lys119 undergoes a major positional shift in the aminoacrylate intermediate where it is no longer required to stabilize the carbanion.

The active site of CBS displays a constellation of conserved interactions common to members of the fold II family of PLP enzymes. PLP is tethered via Lys119 forming an internal aldimine in the resting human enzyme. At the other end of the PLP ring, the side chain of Ser349 is positioned to interact with the pyridinium nitrogen while Asn149 hydrogen bonds with the oxygen. Electrostatic contacts made between conserved threonine residues (Thr257 and Thr260) in a glycine-rich loop and the phosphate group of PLP further lock the cofactor in place. High resolution structures of Drosophila CBS have captured two reaction intermediates, the carbanion and the aminoacrylate species, providing detailed insights into the mechanism of their stabilization.¹⁹⁶ A zwitterionic interaction between the ε-NH₃⁺ group of the lysine, which forms a Schiff base with PLP in the resting enzyme, and the Cα (at 2.0 Å distance) and C4A (at 3.0 Å distance) stabilizes the carbanion intermediate (Chart 7B). In the aminoacrylate intermediate, the ε-NH₃⁺ of the lysine, which is no longer needed to stabilize charge at Cα/C4A, is instead, parked near the phosphate group of PLP, with which it interacts (Chart 7C).

Synthesis of Cys-SSH from cystine represents a β-elimination reaction. It is expected to proceed via a similar reaction sequence up to the formation of the aminoacrylate intermediate, which is accompanied by elimination of the Cys-SSH product. After this point, a transschiffization (rather than a β-replacement) reaction regenerates the resting internal adimine form of the enzyme and releases the eneamine product, which is hydrolyzed in solution to the α-keto acid, pyruvate, and ammonia.

4.1.3. Relative Efficacy of H₂S versus Cys-SSH Synthesis by CBS

Rat¹⁶⁵ and human²²⁶ CBS catalyze the β-elimination of cystine, the oxidized form of cysteine, to form Cys-SSH. The kinetic parameters for human CBS catalyzed Cys-SSH formation are k_cat = 0.11 s⁻¹ and k_cat/K_m = 85 M⁻¹ s⁻¹ at pH 7.4 and 37 °C. Under V_max conditions, the most efficient reaction for H₂S synthesis by human CBS is the β-replacement of cysteine by homocysteine with the following kinetic parameters: k_cat = 19.6 s⁻¹ and k_cat/Km(Cys) = 2882 M⁻¹ s⁻¹ at pH 7.4 and 37 °C. Since the intracellular milieu is reducing and the concentration of cystine is significantly lower than of cysteine, substrate levels regulate H₂S synthesis from cysteine versus Cys-SSH synthesis from cystine. Kinetic simulations of reaction rates at physiologically relevant concentrations of cysteine, cystine, and homocysteine revealed that the contribution of CBS to Cys-SSH synthesis is negligible under these conditions being ∼30 000-fold lower than H₂S synthesis.²²⁶ This analysis suggests that CBS is unlikely to be a significant source of Cys-SSH in cells.

4.2. Regulation of CBS

CBS is a busy hub of regulation, which is fitting since it directs sulfur away from the cycle of an essential amino acid, methionine, to other sulfur metabolites such as cysteine, glutathione (GSH), taurine, and H₂S. Embedded in the CBS structure itself are two domains at the N- and C-termini, which exert their regulation in distinct ways and also appear to impact each other in ways that are poorly understood. In the following section, modulation of CBS activity by its regulatory domains and by posttranslational modifications is discussed.

4.2.1. Heme-Dependent Allosteric Regulation of CBS

The ferric heme in CBS, which is coordinately saturated, is relatively inert to exchange by exogenous ligands.²¹⁴ On the other hand, ferrous CBS binds exogenous ligands such as CO and NO^• with concomitant loss of activity.^217,218 Binding of NO^• to ferrous CBS is accompanied by a shift in the Soret peak from 448 to 390 nm²¹⁸ and leads to a 5-coordinate heme from which both Cys52 and His65 are dissociated (Chart 9). Wild-type CBS exhibits a monophasic binding isotherm for NO^• and a K_D ≤ 0.23 μM.²²⁷ The rate constant for NO^• binding to CBS exhibits a linear dependence on NO^• concentration (8 × 10³ M⁻¹ s⁻¹, pH 7.0 and 25 °C) and is enhanced ∼2-fold in the presence of AdoMet together with a 1.3-fold decrease in k_off. CO displaces the Cys52 ligand and forms a 6-coordinate low-spin ferrous-CO species with a maximum at 420 nm (Chart 8). Binding of NO^• is ∼100-fold faster than of CO, which is limited by the dissociation of Cys52 from the heme iron.²²⁸ Thus, NO^• is presumed to bind by initial displacement of the His65 ligand (Chart 8).²²⁷

Chart 8. Binding of NO^• and CO to Ferrous CBS^a.

^aBinding of NO^• is fast and predicted to occur via displacement of the His65 ligand. Binding of CO is slow and limited by the slow dissociation of the Cys52 ligand.

The affinity of the CBS heme for CO (5-fold) and NO^• (2-fold) is increased in the presence of AdoMet.²²³ Interestingly, deletion of the heme domain reverses the sensitivity of CBS to AdoMet, leading to a 1.5-fold decrease in activity.²¹² The heme ligand mutants, C52A/S, exhibit a similar magnitude of inhibition in the presence of AdoMet.²²⁹ The influence of the C-terminal domain on the heme redox potential (discussed above) and the effect of AdoMet on the affinity of the heme ligands, CO, and NO^• hint at very long-range communication between the N- and C-terminal domains, which are >50 Å apart in the structure of AdoMet-bound CBS.¹⁹⁵

Insights into how changes in the heme domain are communicated over an ∼20 Å distance to the active site have emerged from fluorescence and resonance Raman studies.²³⁰ The ketoenamine tautomer of PLP is key to reactivity since it facilitates the nucleophilic attack by the substrate amino group to form the external aldimine and subsequently stabilizes the carbanion following α-proton abstraction. Changes in the heme ligand environment (e.g., CO binding or heat treatment which displaces the Cys52 ligand)²³¹ shift the PLP equilibrium from the ketoenamine to the enolimine tautomer, in which the proton relocates to the exocyclic oxygen at the C3 atom on the PLP ring. The salt bridge between Cys52 and Arg266 is postulated to be critical for stabilizing the active ketoenamine tautomer.²³⁰ Arg266 resides at one end of an α-helix. At the other end of the same α-helix are two conserved electrostatic interactions between Thr257 and Thr260 and the phosphate group of PLP (Figure 3). Loss of the Cys52-Arg266 salt bridge either via ligand exchange or in the pathogenic R266M mutant, stabilizes the inactive enolimine tautomer.²³⁰ The conservative R226K mutation leads to lengthening of the ferric Fe–S bond and perturbations in the PLP electronic spectrum.²³² The allosteric communication between the heme and PLP sites is bidirectional since the pathogenic T257M mutation promotes loss of the Cys52 ligand in the ferrous state and a concomitant shift to the inactive enolimine tautomer.²³³

4.2.2. AdoMet-Dependent Allosteric Regulation of CBS

The C-terminal regulatory domain imparts both intrasteric and allosteric effects and is responsible for the propensity of the full-length protein to aggregate. Binding of AdoMet increases k_cat ∼2-fold from 2.8 to 5.2 s⁻¹, while deletion of the entire domain increases k_cat 5-fold to 10 s⁻¹ (all values calculated per monomer at 37 °C). The structures of human CBS with and without AdoMet provide molecular insights into the autoinhibitory effect of the C-terminal domain and its alleviation by AdoMet (Figure 2).^194,195 While the catalytic cores in the two structures are virtually identical, the C-terminal domain undergoes a substantial rearrangement. In the absence of AdoMet, the C-terminal domain of each subunit sits on the catalytic core of the adjacent subunit, impeding access to the active site. A combination of hydrophobic interactions between residues in the CBS2 motif (Ile537, Leu540, and Ala544) and the catalytic core (Ile166, Val189, Val206, Leu210, and Ile214) and hydrogen bonding interactions between residues in the CBS1 motif (Thr460, Asn463, Ser466, and Tyr484) and a loop at the active site entrance (Glu201, Asn194, Arg196, and Asp198) lock in this conformation. AdoMet binds in a cleft between the CBS1 and CBS2 domains, which is solvent exposed and is stabilized via hydrophobic interactions and a network of hydrogen bonds. Binding of AdoMet leads to a major structural rearrangement in which the C-terminal domains uncross and dimerize in a head-to-tail fashion on top of the catalytic domain with which all interactions are broken. A flexible linker between the catalytic core and the C-terminal domain (spanning residues 381–411) is critical for mediating the AdoMet-induced conformational change, which leads to unobstructed access to the active site. The structure of human CBS with AdoMet is very similar to that of full-length Drosophila CBS, which is hyperactive in its basal state but does not bind AdoMet.¹⁹⁶

A number of pathogenic mutations in the C-terminal domain (P422L, P427L, I435T, D444N, S466L, and L540Q) render the protein more active than wild-type CBS but insensitive to further activation by AdoMet,^234–236 begging the question as to why they are disease causing. Curiously, a subset of pathogenic mutations in the catalytic core of CBS (A114V, A158V, V168M, A226T, R224H, T262M, I278T, A331V, and T353M) are functionally suppressed by deleting the C-terminal domain (i.e., the last 145 residues)²³⁷ or selecting for mutations in this domain that suppress the most common patient mutation in CBS, I278T.²³⁸ The suppressor mutants are unresponsive to AdoMet suggesting that the mutations stabilize the activated conformation even in the absence of the allosteric ligand.

4.2.3. Regulation of CBS by Covalent Modifications

CBS is regulated by at least three types of covalent modifications: (i) sumoylation, (ii) glutathionylation, and (iii) phosphorylation. CBS is a target of modification by the small ubiquitin-like modifier-1 protein (SUMO-I). A number of proteins belonging to the sumoylation machinery were identified as potential interacting partners of human CBS from a yeast two-hybrid screen and included Pc2, PIAS1, PIAS3, Ubc9, and RanBPM.¹⁸⁶ Of these, Ubc9 is an E2 conjugating enzyme, while PIAS1, PIAS3, and Pc2 are E3 SUMO ligases, which confer target specificity and reaction efficiency. Under in vitro conditions, Pc2 enhances CBS sumoylation, which is correlated with a 70% decrease in activity.²³⁹

The C-terminal regulatory domain is required for the interaction between CBS and the other sumoylation machinery proteins noted above although the modification itself appears to occur in the catalytic core. Mutation of Lys211, embedded in a canonical ΨKXE sumoylation motif and exposed to solvent, leads to loss of sumoylation in vitro, suggesting that this lysine might be tagged by SUMO1. Sumoylation of CBS is correlated with its nuclear localization and can be visualized in cells and in tissue when care is taken to deactivate desumoylases.¹⁸⁶ Interestingly, in porcine brain, sumoylated CBS appears to be the dominant form of the protein. The physiological significance of CBS sumoylation is presently unknown. It could serve to translocate CBS to the nucleus under conditions of stress (e.g., hydrogen peroxide, heat shock, heavy metals, or ethanol treatment) that result in a global increase in sumoylation,^240–242 leading to a local increase in H₂S and/or GSH synthesis assuming that CSE, which is sumoylated in vitro,²³⁹ also relocates under these conditions.

Glutathionylation of CBS is observed under both in vitro conditions and in cultured cells challenged with H₂O₂.²⁴³ This modification leads to a 2–3-fold increase in CBS activity and occurs at Cys346, which resides in the catalytic core near the dimer interface and is not particularly surface exposed. Glutathionylation of CBS renders the protein insensitive to further activation by AdoMet. Cys346 resides in a loop between two α-helices, which are involved in the interface between the catalytic core and the linker region in the (AdoMet-induced) activated conformation of CBS. Hence, modification at Cys346 might stabilize the activated conformation of CBS even in the absence of AdoMet. The functional significance of glutathionylation appears to be to up-regulate transsulfuration flux under oxidizing conditions, which deplete GSH pools, leading to greater synthesis of cysteine. The transsulfuration pathway is known to be an important feeder for cysteine, the limiting reagent for GSH synthesis.^244,245 Glutathionylation of CBS, which is transiently increased in cells in response to oxidative stress, accounts for the increased flux of sulfur through the transsulfuration pathway under these conditions.

In the urothelium, CBS is reportedly phosphorylated at Ser227 and Ser525 in a cGMP/protein kinase G-dependent reaction.²⁴⁶ Phosphorylation was triggered with 8-bromo-cGMP, a stable analogue of cGMP, which caused an ∼2-fold increase in H₂S production in urothelial cell lysates. H₂S production was, however, tested in the presence of a high concentration of cysteine, which is a more effective substrate for CSE than for CBS, both of which are present in the urothelial cells. It was concluded that Ser227 rather than Ser525 is important for the phosphorylation-induced increase in CBS activity based on the responses of transfected cell lines harboring the S227A or S525A mutations.²⁴⁶ While S525A CBS expressing cells exhibited increased H₂S production when treated with 8-bromo cGMP, S227A CBS expressing cells showed very low H₂S synthesis, which was not responsive to 8-bromo cGMP. It should be noted, however, that the S227A mutant itself showed very low activity indicating that the mutant is catalytically compromised even in the absence of phosphorylation. A physiological role for H₂S synthesis in bladder relaxation has been proposed.²⁴⁷

4.3. γ-Cystathionase (CSE)

CSE is the second enzyme in the transsulfuration pathway and is also dependent on the PLP cofactor, for catalysis.^181,182 It catalyzes the γ-elimination of cystathionine to give cysteine, α-ketobutyrate, and ammonia (Chart 9).²⁴⁸ Cysteine synthesis via the transsulfuration pathway is a quantitatively significant source of this amino acid, supplying ∼50% of the cysteine present in the hepatic GSH pool.²⁴⁵ Like CBS, CSE exhibits substantial substrate promiscuity and catalyzes a complex array of H₂S generating reactions involving chemistry at both the β-and γ-carbons of the substrate.²⁴⁹ In addition, CSE catalyzes both Cys-SSH and homocysteine persulfide (Hcy-SSH) synthesis from cystine and homocystine, respectively.^165,226 Mutations in CSE lead to cystathionuria, an autosomal recessive disorder that is generally benign.^250,251 In contrast to CBS, very few pathogenic mutations have been reported in CSE.^251,252

Chart 9. Reactions Catalyzed by CSE^a.

^aThe first reaction is the cleavage of cystathionine to cysteine, α-ketobutyrate (α-KB), and ammonia in the canonical transsulfuration pathway. The next five reactions produce H₂S, while the last two generate the corresponding persulfides from cystine and homocystine.

A significant difference between CBS and CSE is the ability of the latter to form a Schiff base with either cysteine or homocysteine bound to PLP, leading to chemical transformations at either the β- or γ-carbon of the substrate. This renders CSE-catalyzed H₂S synthesis responsive to homocysteine concentrations and H₂S production is predicted to increase between 20- and 200-fold in homocystinuria.²⁴⁹ This explains the clinical observation that homolanthionine (the condensation product of 2 mol of homocysteine) is present in urine of homocystinuric patients with CBS deficiency²⁵³ and emphasizes an underappreciated role for CSE in homocysteine management. Under V_max conditions, the highest k_cat for H₂S generation is for the γ-replacement reaction of homocysteine by a second mole of the same substrate. This is followed by the γ-elimination of H₂S from homocysteine or by the β-replacement of 1 mol of cysteine by another. In addition to H₂S, these reactions produce the novel sulfur metabolites, homolanthionine and lanthionine (Chart 9). At physiologically relevant concentrations of substrates, the β-elimination of cysteine is predicted to be the major H₂S-producing reaction catalyzed by CSE.²⁴⁹

The early part of the CSE-catalyzed reactions proceeds through the same steps as the CBS reaction (Chart 9) leading up to the carbanion intermediate. Thereafter, a second deprotonation at the β-carbon sets up the subsequent γ-elimination of H₂S and formation of the β–γ unsaturated imine intermediate I, which is fully conjugated and can suffer one of two fates (Chart 10). Intermediate I can continue down the γ-elimination path, which involves protonation at the γ-carbon followed by imine hydrolysis to give an eneamino acid. Tautomerization of the eneamino acid to the keto acid (α-ketobutyrate) appears to be enzyme catalyzed as revealed by the stereochemical analysis of the α-ketobutyrate formed by γ-elimination of homoserine. In the γ-replacement path, a second amino acid (e.g., homocysteine or cysteine) adds to the electrophilic γ-carbon in intermediate I forming a condensation product. Following α-carbon protonation and Schiff base exchange with the active site lysine (Lys212 in human CSE), the resting internal aldimine is reformed.

Chart 10. Outline of CSE Mechanism Illustrated for the γ-Elimination of Homocysteine (Red Box) or the γ-Replacement of Homocysteine by a Second mole of the Same Substrate (Blue Box)^a.

^aHcy-S⁻ denotes homocysteine. The first few steps until intermediate I are common to both pathways.

4.3.1. Relative Efficacy of H₂S versus Cys-SSH or Hcy-SSH Synthesis by CSE

Cys-SSH and homocysteine-persulfide (Hcy-SSH) are formed by the CSE-catalyzed β-elimination of cystine and γ-elimination of homocystine, respectively (Chart 9).^165,226 The kinetic parameters for CSE-catalyzed Cys-SSH formation are k_cat = 0.21 s⁻¹ and k_cat/K_m = 1.75 × 10³ M⁻¹ s⁻¹ at pH 7.4 and 37 °C.²²⁶ Due to the insolubility of homocystine, the kinetic parameters for Hcy-SSH synthesis were determined at pH 8.5 and 37 °C and reported to be k_cat = 1.5 s⁻¹ and k_cat/K_m = 221 M⁻¹ s⁻¹. The activity of CSE at pH 7.4 is ∼11-fold lower than at pH 8.5. At physiologically relevant concentrations of substrate, Hcy-SSH synthesis by CSE is predicted to be negligible. H₂S synthesis is estimated to represent ∼98.7% and Cys-SSH only 1.3% of CSE activity. However, a note of caution here is important. The intracellular concentration of cystine is low and not well determined. Allowing for a 25-fold higher hepatic cystine concentration (i.e., 5 μM) than previously reported (i.e., 0.2 μM),¹⁶⁵ Cys-SSH and H₂S synthesis are estimated to represent 33% and 66%, respectively of total CSE activity. CSE rather than CBS is predicted to be the major contributor of hepatic Cys-SSH synthesis at physiological concentrations of cystine, due to the higher protein levels of CSE versus CBS in this tissue.²⁵⁴ The kinetic analysis of H₂S versus Cys-SSH synthesis suggests that, under conditions that lead to increased intracellular cystine levels (e.g., oxidizing conditions), Cys-SSH formation could be elevated.

4.3.2. Structural Organization of CSE

The structures of human CSE with (Figure 4) and without the PLP cofactor and with the suicide inactivator, propargylglycine (PPG) have been reported.²⁵⁵ The protein is a homotetramer (45 kDa monomers) and comprises two domains: a larger PLP domain (residues 9–263) and smaller C-terminal domain (264–401). The PLP is anchored via a Schiff base linkage to Lys212 and Asp187 is involved in an electrostatic interaction with the pyridinium nitrogen (Figure 5A). Tyr114 engages in a π-stacking interaction with the pyridinium ring. Residues contributed from the adjacent subunit also interact with the PLP. Reactions at the γ-carbon require a two-base mechanism (Chart 10); Tyr114 and Lys212 could serve this role.²⁵⁶ On the other hand, reactions at the β-carbon as in the synthesis of H₂S from cysteine, involve a single base, i.e., Lys212. Consistent with this model, mutation of Tyr114 to phenylalanine not only fails to inhibit but, in fact, enhances H₂S synthesis from cysteine 3.6-fold compared to wild-type CSE.²⁵⁷ Unfortunately the canonical cystathionine cleavage activity of this mutant was not reported.

Figure 4.

Structure of human CSE (PDB: 2NMP). Each of the four monomers is shown in a different color, and the three PLPs visible in the structure are in sphere representation.

Figure 5.

Close-up of the active site structure of human CSE. (A). Interactions between PLP (yellow) and residues donated by the two subunits (in green and cyan) are highlighted (PDB: 2NMP). (B). Structure of PPG-inactivated CSE (PDB: 3COG). The coloring is the same is in panel A. PPG is covalently linked to Tyr114 and is shown in pink.

A number of other active site residues are also highly conserved in CSE including Tyr60, Arg62, Thr189, and Arg375. Replacement of any of these residues with alanine leads to loss of H₂S synthesis activity with the exception of the E339A mutant, which exhibits a 3-fold enhancement in the catalytic efficiency of H₂S synthesis from cysteine.²⁵⁷ Based on a structure-based sequence alignment analysis, the residue corresponding to Glu339 in human CSE was predicted to be an important determinant of reaction specificity, i.e., for β- versus γ-elimination.²⁵⁶ Hydrophobic residues at this position are found in enzymes that catalyze β-elimination reactions as in plant and bacterial cystathionine β-lyases. In human CSE the hydrophobic E339Y substitution increased the catalytic efficiency of H₂S synthesis 7-fold.²⁵⁷

Remarkably, the structure of inactivated CSE revealed that the PPG is not bound to PLP; instead, its Cγ is covalently linked to Tyr114 through a vinyl ether linkage (Figure 5B). In fact, the PPG is rotated 180° away from the PLP site, and its amino and carboxyl groups are involved in hydrogen bonding interactions with Arg62 (donated by an adjacent monomer), Glu339, and Arg119. In contrast, the PPG-bound structures of the Trichomonas vaginalis methionine γ-lyase (PDB: 1E5E) and the E. coli CsdB (PDB: 1I29),²⁵⁸ which catalyzes cysteine desulfuration and selenocysteine deselenation, respectively, show that the inhibitor is linked via a Schiff base to the PLP. In methionine γ-lyase, the Cγ of PPG is covalently linked to a tyrosine residue that is homologous to Tyr114 in human CSE. No additional covalent linkages are seen between PPG and CsdB, which is reversibly inhibited by PPG in contrast to irreversible inhibition of CSE and methionine γ-lyase.

4.3.3. Regulation of CSE

CSE levels are markedly reduced in malignant lymphoid cells, which exhibit cysteine auxotrophy.²⁵⁹ Since CSE deficiency can be clinically benign, it has been the subject of limited mechanistic and epidemiological scrutiny, and we understand little about how this enzyme is regulated in comparison to CBS. CSE has two CXXC motifs of which one, Cys₃₀₇-X-X-Cys₃₁₀, is relatively surface exposed. The second motif, Cys₂₅₂-X-X-Cys₂₅₅, is more buried and proximal to the dimer–dimer interface. The potential role of these CXXC motifs in allosteric regulation via redox changes or metal coordination is not known.

Protein kinase G-dependent phosphorylation of CSE at Ser377 has been reported to inhibit H₂S production in the carotid body.²⁶⁰ However, Ser377 is completely buried and it is unclear how this residue can be phosphorylated or how the insertion of a phosphate group in the protein’s interior is stabilized. H₂S synthesis by CSE reportedly increased in the presence of calcium/calmodulin,¹² although other laboratories have not been able to reproduce this observation.²⁶¹ Sumoylation of human CSE has been reported in vitro²³⁹ but the physiological relevance of this modification remains to be assessed.

4.4. Regulation of H₂S Synthesis by the Transsulfuration Pathway

The multitude of reactions catalyzed by CBS and CSE^{165,190,226,249} begs the question as to how the trans-sulfuration pathway responds to cellular demands for cysteine versus H₂S synthesis. At one level, this pathway might be controlled by gene expression, i.e., by the predominance or complete absence of one of the transsulfuration enzymes in some cells. For instance, if CBS expression is turned off, then cystathionine will not be synthesized in that cell type and therefore unavailable as a substrate for CSE (extracellular cystathionine levels are very low). Under these conditions, only the H₂S synthesis reactions of CSE will be catalyzed. On another level, substrate levels control the dominant flux (i.e., serine versus cysteine for CBS and cysteine/homocysteine versus cystathionine for CSE). Finally, allosteric regulation by ligands whose concentrations change transiently in response to stimuli could regulate the dominant metabolic track that CBS and CSE operate on.

All three strategies are involved in regulating flux through the transsulfuration pathway and in triggering metabolic track switching in response to cellular needs.²⁶² Binding of ligands, e.g., CO or NO• to the heme sensor in CBS, can flip the operating preference of the transsulfuration pathway from cysteine to H₂S synthesis (Figure 6).²⁶² Thus, despite the similar catalytic efficiencies for serine (2650 M⁻¹ s⁻¹) and cysteine (2882 M⁻¹ s⁻¹), CBS preferentially synthesizes cystathionine (from serine) over H₂S (from cysteine) under basal conditions due to the higher cellular concentration of serine (∼1–2 mM) than cysteine (∼50–100 μM). The major product of CBS, i.e., cystathionine, is a more efficient substrate for CSE (8200 M⁻¹ s⁻¹) than is cysteine (270 M⁻¹ s⁻¹) or homocysteine (350 M⁻¹ s⁻¹).²⁴⁹ Hence, under basal conditions or upon AdoMet activation of CBS, the predominant flux in the transsulfuration pathway is toward cysteine synthesis (Figure 6, left). In contrast, under conditions that trigger H₂S-signaling such as endoplasmic reticulum stress,²⁶³ CBS is inhibited by CO,²¹⁷ a product of heme oxygenase-1 that is induced under these conditions.²⁶⁴ Furthermore, AdoMet exacerbates CO inhibition of CBS.²²³ In the absence of competition from cystathionine, CSE preferentially uses cysteine (which is ∼100-fold more abundant than homocysteine) to produce H₂S (Figure 6, right). The track-switching model has important implications for sulfur metabolism. For instance, it predicts that H₂S homeostasis will be dysregulated in CBS deficiency-induced hyperhomocysteinemia and suggests that H₂S-dependent signaling cascades are perturbed in complex diseases like cardiovascular, neurodegenerative, neoplastic, and metabolic diseases where compromised endoplasmic reticulum function is a significant factor.²⁶⁵ Finally, there is growing evidence for crosstalk between NO^•, CO, and H₂S signaling pathways,^{262,266–268} but the molecular mechanisms underlying this interconnectedness are far from understood. The transsulfuration pathway represents a platform for the interplay between them via a heme sensor embedded in CBS.

Figure 6.

Heme-regulated switching in the transsulfuration pathway from cysteine (left) to H₂S (right) synthesis.

4.5. Mercaptopyruvate Sulfur Transferase

MST is a sulfurtransferase that is found in the cytoplasm and in the mitochondrion.^189,269 The specific activity of MST is reported to be 3-fold higher in mitochondria than in the cytosol in rat liver.¹⁸⁹ MST catalyzes the transfer of the sulfur atom from 3-mercaptopyruvate, which is derived from cysteine via the action of cysteine aminotransferase (CAT, identical to aspartate aminotransferase), a PLP enzyme that utilizes α-ketoglutarate as a cosubstrate (Chart 11). In the first half reaction, sulfur transfer from mercaptopyruvate to an active site cysteine residue (Cys248 in the human MST sequence) results in a stable Cys-SSH intermediate and formation of pyruvate. In the second half reaction, the outer sulfur from the Cys-SSH intermediate is transferred to a nucleophilic acceptor, which can be a small molecule thiol or the protein, thioredoxin. H₂S is subsequently liberated from the sulfur acceptor (Chart 11B).^270,271

Chart 11. Reaction Catalyzed by MST^a.

^a(A) 3-Mercatopyruvate is synthesized by L-cysteine aminotransferase (CAT), which requires α-ketoglutarate as a cosubstrate. In the first half reaction, MST transfers the sulfur atom from 3-mercaptopyruvate to an active site cysteine forming a Cys-SSH intermediate. (B) In the second half reaction, the outer sulfur from Cys-SSH is transferred to a small molecule thiol acceptor (RSH) or to thioredoxin (Trx) and subsequently released as H₂S. (C) 3-Mercatopyruvate can be synthesized from D-cysteine via the action of D-amino acid oxidase (DAO).

Alternatively, the sulfur transfer can occur to cyanide, generating thiocyanate.²⁷² Based on the k_cat/K_m parameters for sulfur transfer from mercaptopyruvate, the following order for decreasing catalytic efficiency has been reported for various biologically relevant acceptors: thioredoxin ≫ cyanide ≈ dihydrolipoic acid > cysteine > homocysteine > GSH.²⁷¹ Dithiothreitol and mercaptoethanol can serve as surrogate acceptors in vitro. In nature, MST variants fused to thioredoxin are found^273,274 suggesting that thioredoxin is the physiological sulfur acceptor for MST.²⁷⁵ In principle, other proteins could serve as sulfur acceptors by interacting directly with MST; however, the identities of such protein acceptors if they exist, are not known.

An alternative route for 3-mercaptopyruvate synthesis is via the oxidation of D-cysteine by the FAD-dependent peroxisomal enzyme, D-amino acid oxidase (Chart 11C), which was first inferred from studies on cyanide detoxification by rat hepatocytes.²⁷⁶ H₂S production from D-cysteine is repressed by indole 2-carboxylate, an inhibitor of D-amino acid oxidase.²⁷⁷ While the highest MST levels are seen in liver, large intestine and kidney, the highest D-amino acid oxidase levels are seen in kidney and cerebellum. H₂S production from D-cysteine is apparently significantly higher than from L-cysteine in kidney and in cerebellum.²⁷⁷ However, a cellular source for D-cysteine is not known.

MST belongs to the rhodanese superfamily and comprises two domains: an N-terminal domain extending from residues 1–138 and a C-terminal domain extending from residues 165–285 (Figure 7A). A long linker (residues 139–164) connects the two domains. The active site is located in a cleft between the two domains with each contributing residues to it. The structure of human MST has been captured with the persulfide intermediate (Cys248-SSH) and pyruvate (Figure 7B).²⁷¹ A second structure was solved with an apparently unproductive complex in which the 3-mercaptopyruvate substrate forms a mixed disulfide with Cys248.²⁷¹ Mutation of the active site cysteine to serine in rat MST leads to complete loss of activity.²⁷⁸

Figure 7.

Structure and mechanism of human MST. (A) The N and C-terminal domains of human MST (PDB: 4JGT) are shown in blue and green, respectively, and the linker is in red. Pyruvate (cyan) and key active site residues including Cys248 in the Cys-SSH state are shown in stick representation. (B) Close up of the active site captured in a product complex with pyruvate (orange) and Cys-SSH (PDB: 4JGT). Two residues in the catalytic triad (D63 and H74) are donated by the N-terminal domain and are shown in blue. (C) Mechanism of the reaction between the 3-mercaptopyruvate substrate and the Cys248 thiolate.

Two conserved arginine residues, Arg188 and Arg197, which were known from mutagenesis studies to be important for substrate orientation,²⁷⁸ form ionic and hydrogen-bonding interactions with the carboxylate group of the substrate and product (Figure 7B). Arg197 additionally forms a hydrogen bond with the carbonyl group of the substrate/product. A Ser-His-Asp catalytic triad, first noted in the Leishmania MST crystal structure,²⁷⁹ is also present in human MST (Ser250-His74-Asp63). The N-terminal domain contributes Asp63 and His74 to the triad. The Cys248 residue is predominantly deprotonated, acoording to a pK_a of 7.2.²⁷¹ The reaction catalyzed by MST is proposed to involve attack by the thiolate of Cys248 on the sulfur of 3-mercaptopyruvate resulting in the formation of pyruvate and a persulfide (Cys-SSH). This product complex has been captured crystallographically (Figure 7B,C).²⁷¹ The outer sulfur of the persulfide, most likely in the anionic Cys-SS⁻ state, is surrounded by an hexapeptide loop and forms hydrogen bonds or other types of polar interactions with the four backbone amides of Gly249, Ser250, Val252 and Thr253 and with the hydroxyl of the latter. The role of the Ser250 hydroxyl in the catalytic triad might be to interact with the carbonyl group of mercaptopyruvate and facilitate catalysis by polarizing the C═O bond.²⁷⁹ An additional proposal, based on a QM/MM study, suggests that Ser250 is involved in the deprotonation of the 3-mercaptopyruvate substrate thiol and in the stabilization of the transient enolate oxyanion.²⁸⁰ However, mutation of the corresponding serine to alanine or lysine in rat MST decreases k_cat/K_m by a modest ∼4–10-fold, suggesting a relatively minor contribution of this residue to catalysis.²⁷⁸ The mechanism of sulfur transfer is proposed to consist of a stepwise process consisting of deprotonation of the 3-mercaptopyruvate thiol, sulfur atom transfer to the Cys248 thiolate to form nascent persulfenate and pyruvate enolate anions, and protonation of the enolate to the corresponding pyruvate enol, which tautomerizes to the keto form.²⁸⁰ The calculated activation barrier for this process is ∼67 kJ mol⁻¹. In contrast, an alternative process of SH transfer that does not require the initial deprotonation of the 3-mercaptopyruvate thiol has a much higher calculated activation energy barrier, 180 kJ mol⁻¹.²⁸⁰ The electrostatic repulsion between the thiolates of the substrate and of Cys248 during the sulfur atom transfer is proposed to be reduced by interactions of the sulfur with the surrounding amide and hydroxyl groups. Release of pyruvate followed by attack of the thiol acceptor on Cys-SSH moves the sulfane sulfur out of the MST active site, regenerating the resting enzyme (Chart 11). The attack by the thiolate acceptor on the outer sulfur of the persulfide would be favored by the specific geometry of the active site and by the increase in electrophilicity of the outer sulfur caused by hydrogen bonding to the amide and hydroxyl groups of the surrounding loop.²⁸⁰

4.5.1. Regulation of MST

Our understanding of how CAT/MST-dependent H₂S synthesis is regulated is very limited. The active site cysteine in MST is sensitive to oxidation and could potentially be involved in redox regulation of the enzyme.²⁷⁵ Reversible inhibition of rat MST by treatment with stoichiometric quantities of hydrogen peroxide or tetrathionate and rescue of the resulting cysteine sulfenate by reductants such as dithiothreitol or thioredoxin has been reported.²⁷⁵ It is not known whether the active site cysteine residue is also susceptible to overoxidation, which would lead to irreversible inactivation.

A role for redox regulation of rat MST via formation of an intersubunit disulfide bond has been reported. Of the three surface-exposed cysteines in rat CST, two (Cys154 and Cys263) form an intersubunit linkage under oxidizing conditions.^275,281 Reduction of the disulfide bond by thioredoxin increases MST activity 4.6-fold. These cysteines are not conserved in the human protein, which is only observed to exist as a monomer.²⁷¹

Calcium (0–2.9 μM) reportedly inhibits CAT/MST-dependent H₂S production in mouse retinal lysate.²⁸² The inhibitory effect of calcium appears to be on CAT rather than MST since H₂S production from 3-mercaptopyruvate was unaffected by the presence of calcium. The mechanisms by which calcium regulates CAT remain to be elucidated.

4.6. Inhibitors of H₂S Biogenesis

The paucity of specific inhibitors of H₂S-producing enzymes has limited advancements in the field and led to the indiscriminate use of nonspecific PLP enzyme inhibitors such as aminooxyacetic acid and hydroxylamine to “target” H₂S production.¹⁸¹ Alternatively, aspartate, the preferred substrate for CAT/AAT, has been used to indirectly inhibit MST-dependent H₂S production. Historically, the acetylenic substrate analog, propargylglycine, was the first mechanism-based inactivator designed to target CSE.²⁸³ The structure of human CSE revealed that propargylglycine is covalently linked to Tyr114 in the active site but is not attached to the PLP via a Schiff base.²⁵⁵ Propargylglycine exhibits low bioavailability and is typically used at high (1–10 mM) concentrations in cell culture experiments. It also exhibits off-target activity with alanine aminotransferase,²⁸⁴ limiting its utility.

An assessment of the commonly used generic inhibitors confirmed a lack of selectivity for CBS versus CSE with three compounds: hydroxylamine, aminooxyacetic acid, and trifluoroalanine.²⁸⁵ In contrast, two compounds showed selectivity for CSE. Of these, β-cyano-L-alanine (IC₅₀ = 14 μM for human CSE²⁸⁵), a plant-derived neurotoxin, is known to act as a suicide inhibitor of PLP enzymes, which abstract a proton from the β-carbon of substrates, e.g., alanine aminotransferase.^286,287 Injection of β-cyano-L-alanine induces convulsions and rigidity and leads to cystathionuria in rat, indicating CSE inhibition in vivo.²⁸⁸ Aminoethoyxvinylglycine (reported IC₅₀ = 1 μM²⁸⁵ and K_i = 10.5 μM²⁸⁹ for human CSE) is a known inhibitor of ethylene synthesis by the plant enzyme, 1-aminocyclopropane-1-carboxylate synthase²⁹⁰ and of bacterial cystathionine β-lyase.²⁹¹ It is likely to be a more useful reagent in cell culture experiments although information on its bioavailability and toxicity in cell culture and in whole animals is limited. Aminoethoxyvinyl glycine is a slow-binding but reversible inhibitor of cystathionine β-lyase.²⁹¹

More recent efforts to address the gap in selective targeting of H₂S-generating enzymes have involved high throughput screening^292–294 and rational mechanism-based inhibitor design.²⁹⁵ In one study, CBS inhibitors were designed to mimic the product, cystathionine but the α-amino group that forms a Schiff base, was substituted by the heteroatomic functional groups: –NHNH₂, –ONH₂, and –NHOH to form a hydrazone, oxime, and nitrone linkage, respectively, with the PLP.²⁹⁵ The inhibitors showed modest potency against CBS, but their activity against CSE, which also binds cystathionine, was not tested. A high throughput assay of a library of 1900 compounds yielded a O-polymethoxylated flavone, tangeritin, and 1,4-napthaquinone, exhibiting modest IC₅₀ values and are unlikely to be useful inhibitors for cell culture studies.²⁹³ A second high throughput screen of 6491 compounds yielded mostly large flavonoids as CBS inhibitors with a subset showing selectivity against CSE although selectivity against MST was not evaluated. These compounds are unlikely to selectively inhibit CBS in the cell.²⁹⁴ A third screen identified benserazide, a known DOPA decarboxylase inhibitor used for managing Parkinson’s disease, as a CBS inhibitor with modest selectivity against CSE and MST.²⁹²

5. ENZYMATIC H₂S OXIDATION

The toxicity of H₂S is associated with its inhibition of cytochrome c oxidase (half maximal inhibition occurs at ∼0.3 μM in cell extracts and ∼20 μM in intact cells).²⁹⁶ Steady-state H₂S concentrations are maintained at low levels (6–80 nM)^159,179 except in aorta, where the concentration is reportedly ∼20–200-fold higher.¹⁷⁸ Hence, cells have strategies for avoiding H₂S build-up. One such strategy involves the canonical H₂S oxidation pathway, which exists in the mitochondrion and converts H₂S to thiosulfate and sulfate,²⁹⁷ with the product distribution being tissue specific.^298–300 Alternatively, H₂S can be oxidized by globins to thiosulfate and protein-bound hydropolysulfides.^301,302 Much less is known about the enzymology of H₂S oxidation compared to its biogenesis, and in this section, the structures and functions of the human proteins in the mitochondrial H₂S oxidation pathway are discussed, while globins that have H₂S oxidation capacity are covered in section 6.

5.1. Mitochondrial H₂S Oxidation Pathway

The eight-electron oxidation of H₂S to sulfate starts in the mitochondrial matrix and is completed in the intermitochondrial membrane space where sulfite oxidase resides (Figure 8).

Figure 8.

Alternative models describing the organization of the mitochondrial sulfide oxidation pathway. (A) In this model, GSH is the sulfur acceptor from SQR and the product, GSSH, is utilized by either PDO or by rhodanese generating sulfite and thiosulfate, respectively. (B) In this model, sulfite is the sulfur acceptor from SQR and the product, thiosulfate, is utilized by rhodanese to generate GSSH, which is subsequently oxidized by PDO to sulfite. Sulfite is eventually oxidized to sulfate by sulfite oxidase. Q represents coenzyme Q.

The first enzyme in the pathway is sulfide quinone oxidoreductase (SQR), which catalyzes a two-electron oxidation of H₂S to persulfide and reduces coenzyme Q (CoQ). The latter enters the electron transfer chain at the level of complex III, thus connecting H₂S oxidation to ATP and reactive oxygen species formation.^303,304 Beyond this step, there is controversy regarding how the pathway is wired. Depending on whether the persulfide acceptor from SQR is GSH or sulfite, glutathione persulfide (GSSH) or thiosulfate results. GSSH synthesis by SQR would set up a competition between its utilization by persulfide dioxygenase (PDO also referred to as ETHE1), the product of the ethe1 gene, and by rhodanese (Figure 8A). Alternatively, if thiosulfate is the product of SQR, it would first have to undergo a sulfur transfer reaction catalyzed by rhodanese to form GSSH, which would then be oxidized to sulfite by PDO (Figure 8B). Sulfite is eventually oxidized to sulfate via sulfite oxidase.

5.2. Sulfide Quinone Oxidoreductase

SQR is a member of the flavin disulfide reductase superfamily, which catalyzes pyrimidine nucleotide-dependent reduction of substrates.³⁰⁵ Like other family members, SQR houses two redox centers, an active site disulfide and a noncovalent FAD cofactor that relays electrons from H₂S to CoQ. Human SQR is a dimeric³⁰⁶ membrane-anchored protein with a globular domain that faces the mitochondrial matrix. The structure of a mammalian SQR is not yet available, but structures of SQR from Acidianus ambivalens³⁰⁷ (Figure 9) and other bacteria^308,309 provide insights into its likely active site architecture.

Figure 9.

Structure of SQR from A. ambivalens. (A). Structure of an SQR subunit (PDB: 3H8L) in which the covalently bound FAD is shown in stick representation. (B) Close-up of the active site showing FAD and a bridging trisulfide intermediate between Cys350 and Cys178.

SQR is a combination of a sulfurtransferase that generates an active site Cys-SSH intermediate and an oxidoreductase, which oxidizes H₂S as it reduces CoQ with FAD serving as an intermediate electron carrier (Chart 12). The sulfane sulfur from Cys-SSH is transferred to an acceptor, which can be GSH, sulfite, cyanide, or a second equivalent of H₂S.^297,310,311 As discussed in more detail below, the identity of the acceptor under physiological conditions is a subject of controversy. It is noteworthy that the bacterial SQRs do not require sulfur acceptors; instead, they form polysulfide or cyclooctasulfur products. In fact, a trisulfide intermediate is seen in the A. ambivalens SQR structure (Figure 9B).³⁰⁷ Also unlike some bacterial SQRs, the FAD in human SQR is bound noncovalently.³¹⁰ It exhibits maxima at 385 and 450 nm and a promiment shoulder at 473 nm. The two-electron redox potential of the bound FAD is −123 ± 7 mV.³⁰⁶

Chart 12. Overview of the Reaction Catalyzed by SQR^a.

^aIn the sulfurtransferase steps, HS⁻ attacks the disulfide bond in SQR forming a persulfide at Cys379 and the sulfane sulfur (in red) is transferred to an acceptor (GSH, sulfite, sulfide, or cyanide). In the electron transfer steps, two electrons are transferred from HS⁻ through the disulfide to FAD and then to CoQ.

The detailed mechanism of SQR is complex and involves a two-step sulfur transfer and a multistep electron transfer through the protein (Chart 13).^306,310,311 The initial step involves nucleophilic attack of H₂S on the active site disulfide (presumably formed between Cys301 and Cys379 in human SQR) and leads to the formation of Cys379-SSH and Cys201-S⁻ on the re face of FAD. At this step, the formation of an unusually strong charge transfer complex is observed, which exhibits a maximum at 673 nm and extends out to 900 nm.³⁰⁶ The bimolecular rate constant for the formation of the H₂S-dependent charge transfer species is 3.4 × 10⁵ M⁻¹ s⁻¹ at pH 7.4 and 4 °C.³⁰⁶ In the presence of CoQ, the rate constant for the formation of the charge transfer complex increases 2.9-fold indicating that the reaction occurs more efficiently in a ternary complex.

Chart 13. Postulated Reaction Mechanism of SQR^a.

^aNucleophilic attack of HS⁻ on the active site disulfide results in the formation of a Cys-SSH intermediate at Cys379 and a charge transfer (CT) complex, which collapses to a postulated 4a adduct. A sulfur acceptor viz. GSH or sulfite moves the sulfane sulfur from the active site, forming GSSH or thiosulfate (S₂O₃²⁻), respectively, and restoring the active site disulfide. Electron transfer from the reduced flavin, FADH₂, to CoQ completes the catalytic cycle.

A charge transfer complex is also formed in the presence of sulfite, which presumably adds to the disulfide bond forming a sulfocysteine intermediate (Cys379-SSO₃²⁻). However, the bimolecular rate constant for sulfite addition is only 1 × 10² M⁻¹ s⁻¹ at pH 7.4 and 4 °C, which is 3000-fold lower than the k_on for H₂S and is unlikely to be significant except perhaps under pathological conditions when sulfite concentrations are elevated.

A variety of small molecules can accept the sulfane sulfur from the Cys-SSH intermediate in SQR, exhibiting a range of catalytic efficiencies (k_cat/K_M): sulfite (1.7 × 10⁶ M⁻¹ s⁻¹), cyanide (5.1 × 10⁵ M⁻¹ s⁻¹, pH 8.5), H₂S (2.3 × 10⁵ M⁻¹ s⁻¹), and GSH (5.1 × 10³ M⁻¹ s⁻¹) at pH 7.4 and 25 °C unless noted otherwise.^310,311 Cysteine and homocysteine can also serve as sulfane sulfur acceptors and exhibit catalytic efficiencies that are similar to GSH.³¹¹

The controversy regarding the physiological sulfur acceptor (Figure 8) originated from the reported inability of GSH to support SQR activity,³¹⁰ which has since been shown to be incorrect.³¹¹ While GSH is abundant (1–10 mM depending on the cell type), sulfite, which is reactive, is present at low steady-state concentrations. A recent study reported that the intracellular sulfite concentration in rat liver is 9.2 μM based on HPLC analysis of tissue lysates, albeit without mass spectrometric or any other experimental validation of the identity of the compound that comigrated with authentic sulfite.³¹² Based on the k_cat/K_M values discussed above, the estimated apparent k_cat at 10 μM sulfite is 16 versus 36 s⁻¹ at 7 mM GSH, as previously predicted by kinetic simulations.³¹¹ In addition to the high probability that tissue sulfite concentrations were not determined accurately,³¹² the use of sulfite as the primary sulfane sulfur acceptor by SQR runs counter to logic for the following reasons. First, it would appear unlikely that if the SQR active site evolved to use a small molecule like sulfite as a substrate, that it could just coincidentally bind the tripeptide substrate, GSH, too. The crystal structure of human SQR should provide needed insights into this debate. Second, the dependence of the first step in a H₂S oxidation pathway on sulfite, a six-electron oxidized product of the pathway (Figure 8B), would appear to be organizationally illogical and was described as “convoluted” even by its proponents.³¹⁰ Third, thiosulfate is a poor substrate for rhodanese for GSSH synthesis,³¹¹ which is required for continued oxidation since GSSH is the only known substrate for PDO. The only other known thiosulfate sulfurtransferase that catalyzes the conversion of thiosulfate to GSSH is located in the cytoplasm³¹³ and its involvement in the mitochondrial H₂S oxidation pathway would require the unlikely translocation of the reactive GSSH intermediate across compartments. For these reasons, it appears likely that the flow of sulfur through the H₂S oxidation pathway is HS⁻ → GSSH → SO₃²⁻ → SO₄²⁻ + S₂O₃²⁻ as shown in Figure 8A.

5.3. Persulfide Dioxygenase

PDO is a nonheme mononuclear iron-containing mitochondrial matrix protein, which belongs to the metallo β-lactamase superfamily family and has a subunit molecular mass of 28 kDa.³¹⁴ It catalyzes an oxygen-dependent oxidation of GSSH giving sulfite and GSH (eq 20).^297,315,316

$$\text{GSSH}+{\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}\to \text{GSH}+{{\text{SO}}_3}^{2-}+2{\mathrm{H}}^{+}$$ (20)

As isolated, the iron is predominantly in the ferrous oxidation state³¹⁷ and is coordinated by a 2His-1Asp facial triad.^318,319 Mutations in PDO lead to ethylmalonic encephalopathy, which is inherited as an autosomal recessive disease and is associated with severe neurological and gastrointestinal clinical presentations.^315,320,321 Thiosulfate and H₂S accumulate in PDO deficiency, and intriguingly, a tissue-specific reduction in cytochrome c levels is seen in muscle and brain.^315,322,323

In solution, human PDO behaves as a mixture of monomer and dimer^316,319 and, like the Arabidopsis thaliana and bacterial enzymes,^318,324 crystallizes as a dimer.³¹⁹ The structure of human PDO reveals an αββα fold typical of metallo β-lactamase family members (Figure 10A).

Figure 10.

Crystal structure of PDO. (A) Structure of human PDO (PDB: 4CHL) in which the protomers are shown in pink and blue, the iron ion as an orange sphere, and the coordinating histidines and aspartate in stick representation. Cys247 is oxidized as cysteine sulfinate and is also shown in stick representation. (B). Close up of the P. putida PDO (PDB: 4YSL) with bound GSH (green). The cysteine sulfur of GSH is proximal to the iron ion, which is coordinated by a 2His-1Asp facial triad.

His79, His135, and Asp154 coordinate iron together with three water molecules, completing an octahedral coordination. A deep channel exists that leads to the active site and is predicted to be where the GSSH substrate binds. In fact, the interactions between the product, GSH, and the active site residues are visible in the Pseudomonas putida PDO structure where the sulfur of GSH is within 2.5 Å of the iron (Figure 10B).³²⁴ GSH binding displaces a single water ligand. Interestingly, Cys247, located near the surface, is oxidized to cysteine sulfinate in the structures of human and A. thaliana PDO.^318,319 It is not known whether this oxidative modification is autocatalytic and whether it has mechanistic/regulatory import or is silent.³¹⁹

Both GSSH and coenzyme A persulfide serve as substrates; the specific activity of PDO is, however, ∼50-fold higher with GSSH than with coenzyme A persulfide. The k_cat/K_M with GSSH is 1.4 × 10⁵ M⁻¹ s⁻¹ at pH 7.4, 22 °C. There has been limited interrogation of the reaction mechanism of PDO. In analogy with related dioxygenases such as cysteine dioxygenase, a mechanism has been proposed in which binding of GSSH displaces one or more water molecules, simultaneously creating the binding site for O₂ (Chart 14, [1–2]).^316,319 The crystal structure of the P. putida PDO with bound GSH³²⁴ and the structure of human PDO in which GSSH has been modeled,³¹⁹ show monodentate coordination by the sulfur atom. This is distinct from the bidentate coordination seen in cysteine dioxygenase in which the amine and sulfhydryl groups of cysteine serve as ligands to iron.³²⁵ In the mechanism proposed for PDO,^316,319 binding of GSSH and of O₂ leads to an Fe^III-superoxo intermediate (Chart 14, [3]) that is in resonance with an Fe^II-superoxo species in which the sulfane sulfur ligand has a partial cation character [4]. Recombination of the sulfane sulfur radical cation and the Fe^II-superoxo species leads to a cyclic peroxo intermediate [5]. Following O–O bond cleavage, a reactive metal bound oxygen and a sulfoxy cation [6] are formed. Alternatively, cleavage of the Fe–O bond and transfer of the activated oxygen to the sulfoxy sulfur cation gives [7]. Rearrangement of either [6] or [7] gives [8] (Chart 14), which following hydrolysis, yields sulfite.

Chart 14. Postulated Reaction Mechanism of PDO^a.

^aBinding of GSSH to the resting enzyme [1] creates a binding site for O₂ [2]. Formation of a superoxo-Fe^III [3] in resonance with a biradical Fe^II species [4] leads to formation of a cyclic peroxo-Fe^II species [5]. Cleavage of the O–O bond gives [6]. Alternatively, cleavage of the Fe–O bond gives [7]. Binding of H₂O [8], sets up hydrolysis and formation of the product, sulfite. Alternatively, the water that remained coordinated to the metal center could be used for the final hydrolysis step.

Two patient mutations in PDO, T1521 and D196N, have been characterized biochemically.³¹⁶ Both mutations affect the iron content of PDO, decrease its thermal stability, and have smaller effects on either the K_m for GSSH and/or on k_cat. The mutated residues are distal from the active site and the decrease in thermal stability (by 10–15 °C) is likely to be the major biochemical penalty leading to disease.

5.4. Rhodanese

Rhodanese is a sulfurtransferase found in the mitochondrial matrix. Historically, rhodanese was thought to have a role in cyanide detoxification since it can convert thiosulfate and cyanide to thiocyanate (Chart 15A).³²⁶ More recently, its role in the mitochondrial H₂S oxidation pathway has been demonstrated where it preferentially catalyzes thiosulfate synthesis versus utilization (Chart 15, B vs C).^297,311 Elevated rhodanese expression is correlated with lower adiposity and knockout of the rhodanese gene in mouse leads to markedly increased diabetes.³²⁷ Other phenotypic and metabolic expressions of rhodanese deficiency have not yet been described.

Chart 15. Reactions Catalyzed by Rhodanese^a.

^aRhodanese exhibits varied sulfur transferase activities including: (A) thiosulfate:cyanide sulfurtransferase, (B) GSSH:sulfite sulfurtransferase, and (C) thiosulfate:GSH sulfurtransferase. E-SSH denotes the enzyme-bound Cys-SSH intermediate. The red color traces the fate of the sulfur from the donor to the acceptor.

Rhodanese, like MST, belongs to the sulfurtransferase superfamily, characterized by a “rhodanese” domain fold with an α/β topology named after this protein, which is present in a single copy, in tandem repeats or fused with other proteins in members of this family.^328,329 Human rhodanese is a monomeric protein with a molecular mass of 33 kDa.³³⁰ Two polymorphic variants of rhodanese have been described, which lead to E102D and P285A substitutions.³³¹ The Glu102 residue is located at the entrance to the active site pocket and is ∼19 Å away from the catalytic cysteine, Cys257. Pro285 is surface exposed and ∼17 Å away from the active site. Interestingly, both variants exhibit greater thermal stability than wild-type rhodanese.³³⁰

The structure of bovine liver rhodanese³³² serves as a useful model for the human protein with which it shares 89% sequence identity (Figure 11). The structure comprises two globular domains of approximately equal length and an active site that is housed in a cleft between the two domains. The C-terminal domain provides the catalytic cysteine and a mixture of hydrophilic and hydrophobic residues wall in the active site. The reaction mechanism of rhodanese involves an active site Cys-SSH intermediate from which the sulfane sulfur is transferred to an acceptor. In the bovine rhodanese structure, the Cys-SSH intermediate is stabilized by hydrogen bonds from the hydroxyl group of Thr252 and the backbone amides of Arg248, Lys249, and Val251.³³² The catalytic triad present in MST is absent in rhodanese. This coincides with its preferential substrates being predominantly deprotonated at physiological pH.²⁸⁰

Figure 11.

Structure of bovine rhodanese (PDB: 1RHD). The N- and C-terminal domains are shown in green and blue with the intervening linker in pink. A Cys-SSH intermediate is stabilized at Cys247 in the active site, via hydrogen bonding interactions with neighboring residues.

Like MST, the reaction catalyzed by rhodanese involves two sulfur transfer reactions: from the sulfur donor to the active site cysteine and from Cys-SSH to the sulfur acceptor (Chart 15). The various sulfur transfer reactions catalyzed by wild-type rhodanese and its polymorphic variants have been characterized.^311,330 Human rhodanese preferentially catalyzes sulfur transfer in the direction of GSSH → S₂O₃²⁻ (k_cat/K_m(sulfite) = 6.5 × 10⁶ M⁻¹ s⁻¹ at pH 7.4, 37 °C) versus in the reverse direction, S₂O₃²⁻ → GSSH (k_cat/K_m(GSH) = 0.03 × 10³ M⁻¹ s⁻¹ at pH 7.4, 37 °C). Based on these values, rhodanese exhibits an estimated 217 000-fold discrimination against utilization of thiosulfate versus GSSH as a sulfur donor. The K_m values for GSSH (450 μM) and sulfite (60 μM) in the GSSH → S₂O₃²⁻ sulfurtransfer reaction are lower than the K_m values for the substrates, thiosulfate (340 μM) and GSH (21 mM), in the S₂O₃²⁻ → GSSH direction. Like wild-type rhodanese, the polymorphic variants also exhibit a marked preference for making rather than utilizing thiosulfate and exhibit comparable catalytic efficiencies for this reaction (k_cat/K_m(sulfite) = 1.5–4.2 × 10³ M⁻¹ s⁻¹ at pH 7.4, 37 °C).³³⁰

Interestingly, cysteine and homocysteine can replace GSH as sulfur acceptors in the S₂O₃²⁻ → GSSH sulfur transfer reaction with catalytic efficiencies (k_cat/K_{m(Hcy or Cys)} ≈ 0.4 × 10³ M⁻¹ s⁻¹ at pH 7.4, 37 °C) that are ∼13-fold higher than with GSH.³¹¹ However, the concentration of these amino acids is low in most tissues, and their high K_m values (∼21 mM each) make them unlikely substrates for the reverse reaction under physiological conditions compared to GSH. In some tissues like kidney, which has high cysteine,³³³ or under pathological conditions like homocystinuria, which is characterized by elevated homocysteine,¹⁹³ these sulfur containing amino acids might become relevant substrates, albeit in the less favorable reverse sulfur transfer reaction from thiosulfate.

The thiosulfate:cyanide transfer kinetics of wild-type rhodanese are characterized by relatively low catalytic efficiency (k_cat/K_m(KCN) = 31 × 10³ M⁻¹ s⁻¹ at pH 7.4, 25 °C) and high K_m for cyanide (29 mM) making a role for rhodanese in cyanide detoxification unlikely. Interestingly, the E102D mutant shows higher efficiency (k_cat/K_m(KCN) = 534 × 10³ M⁻¹ s⁻¹) while the P285A mutant shows similar efficiency (48 × 10³ M⁻¹ s⁻¹) as the wild-type protein in the cyanide detoxification assay.

5.5. Sulfite Oxidase

The nearly ubiquitous presence and conserved architecture of sulfite oxidases is consistent with the evolutionarily ancient role of this protein in protecting against sulfite-induced damage.³³⁴ Sulfite oxidase is a molybdenum containing protein, which catalyzes the two-electron oxidation of sulfite to sulfate in which water serves as the oxygen source (eq 21).

$${{\text{SO}}_3}^{2-}+{\mathrm{H}}_2\mathrm{O}\to {{\text{SO}}_4}^{2-}+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}$$ (21)

In humans, sulfite oxidase is a soluble enzyme found in the mitochondrial intermembrane space. Electrons from the sulfite oxidation reaction are passed via a heme cofactor found in vertebrate sulfite oxidases to cytochrome c and from there to complex IV. The important role of sulfite oxidase in detoxifying sulfite is borne out by its presence in the peroxisomal compartment in plant cells where it functions to remove toxic sulfite derived from atmospheric sulfur dioxide or from catabolism of sulfur containing amino acids, rather than a role in sulfur assimilation in the chloroplast.³³⁵ Sulfite oxidase deficiency is an autosomal recessive disorder that presents with severe neonatal neurological problems.³³⁶ It can result from defects in the synthesis of the molybdopterin cofactor or from mutations in the gene encoding sulfite oxidase itself.

Sulfite oxidase is a homodimer with a subunit molecular mass of 52 kDa and contains a heme b cofactor housed in the N-terminal domain that is connected via a flexible linker to the central molybdopterin-binding domain, which in turn is followed by the C-terminal dimerization domain. The structure of chicken sulfite oxidase³³⁷ (Figure 12) serves as a useful model for the human protein with which it shares 68% sequence identity. The 5-coordinate molybdenum center has square pyramidal geometry (Chart 16A). Of the three sulfur ligands, two are derived from the dithiolene group of the molybdopterin cofactor, while the third is donated by Cys207 (human sequence numbering). The remaining coordination sites are occupied by equatorial and apical oxo ligands. The substrate-binding pocket in chicken sulfite oxidase comprises Arg138, Arg190, and Arg450 in addition to Tyr322 and Trp204. A pathogenic mutation in a patient with severe sulfite oxidase deficiency has been mapped to Arg160, which corresponds to Arg138 in the chicken sequence.³³⁸

Figure 12.

Structure of dimeric chicken sulfite oxidase (PDB: 1SOX). The two subunits are shown in blue and magenta, respectively, and the heme (orange) and molybdopterin (MPT, cyan) cofactors are in sphere representation.

Chart 16. Molybdopterin Cofactor in Sulfite Oxidase and Redox Changes during the Catalytic Cycle^a.

^a(A) Attack of sulfite on an oxo/hydroxyl ligand reduces the molybdenum ion. (B) The reaction cycle of sulfite oxidase involves an initial two-electron reduction of the molybdenum center, which is subsequently oxidized in two one-electron steps via intramolecular electron transfer to the heme. The latter in turn, transfers electrons to the heme in cytochrome c (cyt c) in an intermolecular process.

The catalytic cycle of sulfite oxidase comprises reductive and oxidative half reactions. Sulfite binds transiently to an equatorial oxo/hydroxyl ligand and reduces the molybdenum center to Mo^IV (Chart 16A). Sulfite is released as sulfate following hydrolysis. The k_cat/K_m(sulfite) is 4.7 × 10⁶ M⁻¹ s⁻¹ at pH 7.5 and 25 °C.³³⁹ In the oxidative cycle, sequential one-electron transfers occur from Mo^IV to the exogenous electron acceptor, i.e. heme iron in cytochrome c, via the heme b cofactor in sulfite oxidase (Chart 16B). The catalytic mechanism of sulfite oxidase has been extensively characterized by spectroscopic and rapid reaction kinetic methods combined with mutagenesis studies and has been reviewed recently.³³⁴

6. CHEMICAL BIOLOGY OF H₂S

Notwithstanding the growing body of evidence for the biological roles of H₂S, the gap between the physiological effects of H₂S and its mechanism of action remains large. Based on chemical principles, H₂S reactivity can be categorized into three reaction groups: (i) binding to and/or redox reactions with metal centers, (ii) cross-talk with and scavenging of reactive oxygen (ROS) and reactive nitrogen species (RNS), and (iii) oxidative modification of protein cysteines to form the corresponding persulfides (Figure 13).^340–342 In this section, we provide an overview of the reactions grouped in (i) and (ii) while protein persulfidation is discussed in section 7.

Figure 13.

Biological reactivity of H₂S. H₂S can react directly with oxidants such as superoxide, HOCl, and ONOO⁻. It can also react with NO^• and S-nitrosothiols leading to the formation of other signaling species (right arrow). Metal centers in proteins can bind H₂S for delivery to specific targets, be reduced by H₂S, or catalyze sulfide oxidation chemistry (left arrow). H₂S is also involved in the modification of protein cysteine residues leading to persulfide (Cys-SSH) formation (central arrow).

6.1. Interaction of H₂S with Metal Centers

Different possilbilites exist for the interaction between H₂S and a metal center. First, H₂S (or HS⁻) can coordinate the metal ion. Second, H₂S can reduce the metal center, concomitantly forming HS^• and other downstream sulfur oxidation products. Third, H₂S can modify heme porphyrins covalently.

The first described biological effect of H₂S identified in 1929 by Keilin was its toxicity, which was ascribed to inhibition of respiration by targeting cytochrome c oxidase (CcO).³⁴³ The reaction with H₂S was later used to stabilize cytochrome c oxidase for its spectral characterization.^344,345 CcO is the final acceptor in the mitochondrial electron transport chain, which uses electrons delivered by cytochrome c to reduce oxygen to water.³⁴⁶ It contains two copper centers (Cu_A and Cu_B) and two heme iron centers (a and a₃).^347–349 Oxygen binds to ferrous heme a₃. NO^• and CO also inhibit the enzyme reversibly.^350–352 Inhibition of CcO by H₂S is almost as strong as with CN⁻, with K_i of ∼0.2 μM,³⁵³ and it is noncompetitive with respect to both cytochrome c and oxygen. The work of Nicholls and colleagues was instrumental in pointing out that, in addition to inhibiting CcO, H₂S might also serve as a substrate/electron donor.^353–356 They observed that the initial product of H₂S/CcO (aa₃) interaction is not an inhibited form of the enzyme and that >1 mol of sulfide/mol CcO was required for full inhibition.³⁵⁵ Binding of H₂S to catalytically active CcO (k_on = 1.5 × 10⁴ M⁻¹ s⁻¹, k_off = 6 × 10⁻⁴ s⁻¹, and K_D = 4 × 10⁻⁸ M⁻¹) is much tighter than the binding of ligands such as azide or fluoride (Chart 17).

Chart 17. Interaction of H₂S with Cytochrome c Oxidase^a.

^a(A) At low concentrations, H₂S binds to ferric heme a₃ and reduces it with concomitant formation of HS^• which can react with either another molecule of H₂S or with oxygen. Reduction of heme a₃ leads to an increase in oxygen consumption. (B) At moderate levels, H₂S interacts with the Cu_B⁺ center forming a stable Cu_B–SH⁻ complex, which is difficult to oxidize. Cu_B⁺ is formed by electron transfer from ferrous heme a₃ or by a direct reduction by H₂S. At higher levels of H₂S, the Cu_B–SH⁻ complex induces a conformational change and causes further binding of H₂S to ferric heme a₃. (C) Alternatively, H₂S can reduce cytochrome c thereby increasing CcO reduction and respiration.

A generalized mechanism has been proposed to explain the interaction of H₂S with CcO.^356,357 At low levels (1:1 ratio of H₂S:CcO), H₂S reduces ferric heme a₃ with concomitant oxidation to HS^•. The reduction of this heme iron by H₂S is thermodynamically unfavorable, and it is likely that HS^• removal, e.g., by reaction with HS⁻ to form H₂S₂^•− (eq 9), pulls the reduction in the forward direction. Heme iron reduction promotes oxygen binding and reduction, explaining why low concentrations of H₂S stimulate respiration.^355,358 Alternatively, HS^• can react with oxygen to form HSOO^• (eq 10) further contributing to oxygen consumption. At moderate concentrations (2–3 fold excess of H₂S), H₂S coordinates to the Cu_B center forming a stable Cu–SH₂ complex as documented by EPR. It is likely that HS⁻ reduces Cu_B^II first and then coordinates to Cu_B^I forming a stable complex that is difficult to reoxidize. In the presence of a large excess of H₂S, HS⁻ binds to ferric heme a₃, in a process that is likely aided by a conformational change caused by HS⁻ binding to Cu_B^I (Chart 17).

Using a synthetic CcO model system, a similar behavior was observed, i.e., that at low H₂S concentration, the ferric iron center was reduced but stable H₂S binding was not observed.³⁵⁹ They also noted that cytochrome c can be reduced at low H₂S concentrations, thus injecting more reducing equivalents into the electron transfer chain and stimulating oxygen consumption (Chart 17).³⁵⁹ Second order rate constants for H₂S-induced cytochrome c reduction observed under aerobic (81 ± 5 M⁻¹ s⁻¹) and anaerobic (480 ± 2 M⁻¹ s⁻¹) conditions differed ∼6-fold at 25 °C.⁹⁷

Another well-documented reaction of H₂S is its reaction with hemoglobin (Hb) and myoglobin (Mb), known since the 19th Century,^360–362 when a green compound was reported to form upon treatment of oxy-Hb or oxy-Mb with H₂S.³⁶³ Although H₂S poisoning resulting in sulfhemoglobinemia is rare, misuse of sulfadrugs (sulfonamides) can lead to “green blood”.³⁶⁴ Sulfhemoglobin (λ_max ∼ 618 nm), results from covalent addition of sulfur to a double bond in one of the pyrrole rings^365–369 leading to the formation of a chlorin type heme (Chart 18A).³⁷⁰ This covalent modification results in significant delocalization of π electron density away from the iron, reducing its affinity for O₂ (∼2500-fold in Mb and ∼135-fold in Hb).³⁶⁹ Mb can be recovered by treating sulfmyoglobin with azide or cyanide, which probably react with the sulfur inserted in the pyrrole ring.³⁷¹

Chart 18. Sulfhemoglobin Formation^a.

^a(A) One of the proposed structures of sulfheme. (B) The mechanism of sulfheme formation is not fully understood (as denoted by the question marks) but starts with compound I or II reacting with H₂S and results in sulfur being incorporated into the porphyrin ring.

Despite extensive studies, the actual mechanism of sulfheme formation is still unclear. It is postulated to involve the formation of an oxoferryl [Fe^IV═O Por^•+] or [Fe^IV═O] intermediate (Chart 18B).³⁷² Sulfmyoglobin is formed stoichiometrically in the reaction between H₂S and metmyoglobin peroxide.^371,373 Sulfcatalase is formed in the reaction between compound II catalase and H₂S.³⁷¹ H₂S inhibits the heme in catalase in two ways: by irreversibly modifying the porphyrin and by reversibly ligating to the iron.³⁷¹ Like catalase, lactoperoxidase compound II reacts with H₂S to form sulflactoperoxidase.³⁷⁴ Studies on myeloperoxidase, a hemeprotein that produces hypochlorous acid and other oxidants for killing pathogens,³⁷⁵ indicate that H₂S is a potent inhibitor (IC₅₀ = 1 μM). H₂S exhibits high bimolecular rate constants for reactions with compound I (1.1 × 10⁶ M⁻¹ s⁻¹) and compound II (2 × 10⁵ M⁻¹ s⁻¹).³⁷⁶ Surprisingly, the reaction of H₂S with Fe^III, Fe^II, compound I, or compound II resulted in the formation of a ferrous–H₂S complex.

The activity of soluble guanylate cyclase (sGC) can also be modulated by H₂S. Essential for NO^• sensing, ferric sGC is reduced by HS⁻, which in turn facilitates NO^• binding and activation of cyclic guanosine monophosphate synthesis.³⁷⁷

6.1.1. Catalytic H₂S Oxidation by Methemoglobin, Myoglobin, and Neuroglobin

The ability of free hemin to oxidize H₂S to thiosulfate was also known for a long time.³⁷⁸ In fact, the design of a H₂S sensor is based on its affinity for ferricmyoglobin (Fe^III–Mb or metmyoglobin).³⁷⁹ These observations presaged the discovery of catalytic H₂S oxidation by methemoglobin (Fe^III–Hb)³⁰¹ and Fe^III–Mb³⁰² to a mixture of thiosulfate and iron-bound hydropolysulfides (Chart 19). Red blood cells lack mitochondria and, therefore, do not have the canonical H₂S oxidation pathway. Yet, these cells have MST and, therefore, the capacity to make H₂S^301,380 raising the question as to whether alternative mechanisms exist for disposing H₂S in these and other cells. The search for an answer to this question resulted in the discovery of catalytic H₂S oxidation by globins containing iron in the ferric oxidation state, as described below.

Chart 19. Minimal Reaction Mechanism for Ferric Globin-Dependent Sulfide Oxidation to Thiosulfate and Iron-Bound Hydropolysulfides^a.

^aDetails of the oxidation chemistry that lead to the products have been omitted for clarity.

Binding of H₂S to Fe^III–Hb or Fe^III-Mb to give the corresponding HS⁻–Fe^III species (Chart 19, [1]) is readily monitored by a shift in the Soret maximum from 405 → 423 nm in Hb³⁰¹ and from 409 →428 nm in Mb.³⁰² A concomitant resolution of the α/β bands at 577 and 541 nm in Hb and 578 and 545 nm in Mb is observed. The bimolecular rate constant for H₂S binding to Fe^III–Hb is 3.2 × 10³ M⁻¹ s⁻¹, the k_off is 0.05 s⁻¹ and K_D is 17 μM at pH 7.4, 37 °C. The corresponding values for Mb are k_on = 1.6 × 10⁴ M⁻¹ s⁻¹, k_off = 1.6 s⁻¹, and K_D = 96 μM. The K_D values represent upper limits since the rate constant for H₂S binding to Fe^III–Hb and Fe^III–Mb increases with decreasing pH and only ∼20% of the dissolved sulfide exists as H₂S at pH 7.4, where the measurements were made. Binding of H₂S to sperm whale Fe^III–Mb has also been reported (K_D = 18.5 μM at pH 7.5 and 20 °C).³⁸¹

Binding of H₂S to Fe^III–Hb results in the conversion of a high-spin g = 5.83 signal to a low-spin rhombic signal with g values of 2.51, 2.25, and 1.86.³⁰¹ Similar changes are observed with Fe^III–Mb, which converts from a high-spin g = 5.92 signal to a low-spin g = 2.57, 2.27, and 1.85 signal.³⁰² Signal integration reveals less than stoichiometric spin concentration associated with the low-spin species indicating the presence of spin silent (diamagnetic and/or an integer spin species) intermediate(s) even at the earliest time point at which the spectra were recorded following H₂S addition. Computational modeling suggested that the electronic structure of the S = ⁵/₂ species can be described as a resonance hybrid of high-spin Fe^III-SH⁻ and high-spin Fe^II–^•SH. However, this model is 116 kJ/mol higher in energy than the S = ¹/₂ model. Therefore, the initial intermediate is best described as Fe^III–SH⁻ with probably a small contribution of the HS^• coordinated structure.

The HS⁻–Fe^III species has been characterized by multiple approaches. These include resonance Raman spectroscopy and X-ray absorption spectroscopy, which reveal the initial formation of a 6-coordinate low-spin ferric species. The resonance Raman spectrum reveals the subsequent formation of high-spin ferrous species, albeit it is unclear whether the signal represents one or more likely, multiple intermediates. Coordination of H₂S to Mb was observed by ultrahigh resolution ESI time-of-flight cryo-MS under anaerobic conditions even when H₂S was in excess. Additional evidence for the initially formed ferric sulfide species comes from the X-ray structure of H₂S treated with hemoglobin solved at 1.8 Å resolution (Figure 14A).³⁸² The structure shows extra density above the iron on the distal side of the heme, which was assigned as sulfide based on the sulfur anomalous difference map. The Fe–S distance is 2.2 Å in both the α and β-subunits of hemoglobin and the HS⁻–Fe^III intermediate is stabilized via hydrogen bonding to a histidine. Interestingly, a second sulfide was captured at the surface of the α-subunit, at the mouth of the PHE path, previously proposed to serve as an entry/exit channel for iron ligands.

Figure 14.

Binding of H₂S to protein metal centers. (A) Structure of the α subunit of human hemoglobin showing HS⁻ bound at the entry/exit point of the so-called Phe path that leads to the distal face of the heme. A second sulfide is bound to the heme iron (PDB: 5UCU). (B) Close up of the heme in hemoglobin I from Lucina pectinata with HS⁻ coordinated to the iron ion (PDB: 1MOH). (C) Structure of hemoglobin-like protein C1 from Riftia pachyptila (PDB: 1YHU) with Zn²⁺ shown in blue and iron hemes in purple.

Bound, H₂S probably exists in equilibrium with [Fe^II–HS^•] which could react with another HS⁻ to form H₂S₂^•−. Coordinated H₂S₂^•− could react further with HS⁻ and oxygen leading to the propagation of hydropolysulfide chain coordinated to Fe²⁺ or to formation of thiosulfate (Chart 19). In principle, the ferrous heme-bound HS^• radical could also react with O₂ to form HSO₂^• (see section 2).

Under anaerobic conditions, binding of 1 equivalent of H₂S to ferric iron is observed. However, under aerobic conditions, net consumption of H₂S is seen with formation of thiosulfate and hydropolysulfides, which remain iron bound. In the proposed mechanism, the second intermediate is an iron-bound hydrodisulfide (Chart 19, [2]), which has been observed by cryo-MS on Fe^III–Mb samples treated with Na₂S. Exposure of Fe^III–Hb to Na₂S₂ results in a shift in the Soret peak from 405 to 421 nm and in the appearance of α/β peaks at 575 and 543 nm, which is similar to the spectrum of HS⁻–Fe^III.³⁸² Under aerobic conditions, thiosulfate is formed from Na₂S₂ in the presence of Fe^III–Hb.

Since the intracellular milieu is reducing, the fate of the iron-bound hydropolysulfides in the presence of physiologically relevant reductants is a pertinent issue. In the presence of GSH, the iron-bound hydropolysulfides are unstable, and GSSH, GSSG, and H₂S products are observed.³⁸² If formed, hydropolysulfides generated via globin-dependent oxidation are unlikely to be stable in the cell and would be converted to GSSH or other persulfides.

The catalytic nature of H₂S oxidation by Hb and Mb at the expense of oxygen is evident from the stoichiometric excess of products formed over heme iron concentration. Furthermore, exposure of sulfide-treated Fe^III–Hb to NADPH/flavin oxidoreductases led to the formation of O₂–Fe^II-Hb with a shift in the Soret peak from 423 to 415 nm. Collectively, these results establish that (i) Fe^III–Hb and Fe^III–Mb can catalyze multiple rounds of sulfide oxidation, and (ii) the O₂-liganded globin can be reformed in the presence of reductases like methemoglobin reductase.

Unlike Hb and Mb in which the distal heme site is available for binding exogenous ligands, the heme in neuroglobin has bis-histidine coordination.³⁸³ The function of neuroglobin, which is highly expressed in neuronal tissues and in some metabolically active tissues, is not known.³⁸⁴ Despite the coordinately saturated iron site, ferrous neuroglobin can bind O₂, CO and NO^•.^385–388 The presence of the distal histidine ligand does in fact mute the reactivity of ferric neuroglobin (Fe^III–Nb) toward H₂S and leads to slow reduction to the ferrous state and to inefficient formation of thiosulfate and hydropolysulfides.³⁸⁹ In the presence of sulfide, the Soret peak of Fe^III–Nb shifts from 412 to 415 nm and the α/β bands are broad and centered at 540 nm with a 575 nm shoulder. It is unclear what this spectral change represents, but it is likely to be a mixture of species as also suggested by EPR and resonance Raman spectroscopy. The EXAFS data do not show evidence for an iron–sulfur bond in sulfide-treated neuroglobin, indicating that the sulfide oxidation products are formed even in the absence of direct coordination to iron. The k_on for the interaction of sulfide with Fe^III–Nb is 13.8 M⁻¹ s⁻¹ at pH 7.4 and 25 °C, which is significantly smaller than the values for Fe^III–Hb and Fe^III–Mb. The k_off and K_D for the interaction of sulfide with Fe^III–Nb are 5 × 10⁻³ s⁻¹ and 370 μM, respectively. As expected, the H64A mutation of the distal histidine residue allows direct binding of sulfide as confirmed by EXAFS analysis, increases the rate constant for sulfide binding 4000-fold, and supports active oxidation of sulfide to thiosulfate and protein bound hydropolysulfides. A rich array of oxidation products were identified with the H64A mutant using cryo-MS including hydropolysulfides with 2–6 sulfur atoms and variously oxygenated derivatives in addition to thiosulfate and sulfate.³⁸⁹

Collectively, the studies on the globins reveal the potential for ferric-iron dependent sulfide oxidation chemistry, whose relative importance in the cell awaits evaluation. An open ligation site promotes sulfide coordination and oxidation chemistry, and in its absence, iron reduction is supported. The relatively low steady-state concentration of H₂S likely reduces the prevalence of reactions between sulfide and heme or nonheme iron (or other metalloproteins) except in special cases like red blood cells where Fe^III–Hb represents 1–3% of total hemoglobin, whose concentration is high (∼5 mM).

6.1.2. Binding and Transport of H₂S by Globins

Specialized hemoglobins that transport H₂S are found in organisms that are adapted to life in sulfide-rich environments. The best-studied example of such a Hb is from the clam, Lucina pectinata (Figure 14B),^{381,390–403} which lives in H₂S-rich waters. The monomeric L. pectinata Hb hemoglobin I (HbI) transports H₂S to symbiotic bacteria, which assimilate it and provide the host with a source of organic sulfur. HbI exhibits a high association constant (k_on = 2.3 × 10⁵ M⁻¹ s⁻¹) and an unusually low dissociation constant (k_off = 0.22 × 10⁻³ s⁻¹) for H₂S,³⁹⁰ which suggests the stabilization of distal sulfide ligand by the active site.^{391,395,396,398,400,401} In human Hb, a histidine residue hydrogen bonds with the iron-bound sulfide. The corresponding residue in HbI is a glutamine, which has a flexible side chain. Mutation of the glutamine residue in HbI to valine precludes heme reduction, while mutation to histidine promotes formation of sulfhemoglobin. Another difference from human Hb, is the presence of phenylalanines in HbI that form a hydrophobic pocket around the sulfide.⁴⁰¹ It is unclear whether H₂S release occurs via slow dissociation or by heme iron reduction. Introduction of positively charged substituents on the porphyrin ring changes the reactivity of metal porphyrins from simple binding of H₂S to catalytic oxidation of H₂S.⁴⁰⁴ Hence, at low concentrations, H₂S release could be due to its dissociation from the heme iron, while at high concentrations, heme reduction and H₂S/hydropolysulfide delivery might predominate.⁴⁰⁵

Another sulfide-adapted organism, the giant tubeworm, Riftia pachyptila, lives in deep-sea hydrothermal vents in symbiotic relationship with sulfide-oxidizing bacteria that need both H₂S and O₂.^406–408 The Riftia hemoglobins are large proteins with a molecular mass of ∼3500 kDa (Figure 14C). Binding of H₂S and O₂ occurs at separate sites. While O₂ binds at the heme iron site, it is unclear where H₂S binds. The protein contains 12 Zn²⁺ ions, which have been suggested as potential sites for H₂S binding.⁴⁰⁸

6.1.3. Interaction of H₂S with Zn^II-Containing Proteins

The interaction of H₂S with Zn^II-containing proteins is poorly studied. It is reported that H₂S represses androgen receptor transactivation by targeting the second zinc-finger module.⁴⁰⁹ Phosphodiesterase 5, a Zn^II-containing enzyme, is inhibited by nanomolar H₂S concentrations.⁴¹⁰ Zinc–hydrogensulfido complexes are not easy to prepare and isolate and require bulky apolar ligands.⁴¹¹ The synthesis of a stable zinc hydrogensulfido complex with the tris(2-pyridylmethyl)amine ligand has been reported.⁴¹² H₂S was released from the complex in acidic medium or transferred to a zinc center with higher affinity via intermediate formation of a μ-sulfido dinuclear species. The ability of Zn^II to coordinate HS⁻ was reported to depend on the ability of the HS⁻ ligand to form hydrogen bonds.⁴¹³ Chemical modifications on the ligand that precluded hydrogen bonding with HS⁻ resulted in decomposition of the complex and ZnS precipitation. This study highlighted the importance of the second coordination sphere in stabilizing the Zn-HS⁻ adduct, suggesting that the protein environment could do the same.

6.2. Interaction of H₂S with ROS and Other Biologically Relevant Oxidants

Being at the lowest oxidation state of −2, the sulfur in H₂S can only undergo oxidation. Oxidation leads to sulfate (SO₄²⁻), sulfite (SO₃²⁻), thiosulfate (S₂O₃²⁻), persulfides (RSS⁻), organic (RSS_nSR) and inorganic (HSS_nSR) polysulfides, and elemental sulfur (S_n). The direct reaction of H₂S with O₂ is thermodynamically disfavored (see section 2).^63,414 Given the high one-electron reduction potential (E°′(HS^•, 2H⁺/H₂S) = +0.91–0.94),^64,65 only relatively strong one-electron oxidants can oxidize H₂S to HS^•, with further reaction of HS^• providing an additional driving force. Indeed, several biologically relevant oxidants can support the one-electron oxidation of H₂S, such as hydroxyl radical,^415,416 carbonate radical,⁶⁵ nitrogen dioxide,⁶¹ and myeloperoxidase oxoferryl compounds I and II;³⁷⁶ the rate constants of these reactions are shown in Table 2. The list of one-electron oxidants that can oxidize H₂S can probably be extended to peroxyl and phenoxyl radicals as well as to other metal centers (see section 6.1). The superoxide radical can also oxidize H₂S.⁹⁷ The apparent rate constants at pH 7.4 vary depending on the oxidant and are similar to those reported for cysteine and GSH.⁶¹ Mixtures of polysulfides and polysulfide radical anions (S₂^•− and S₃^•−) are observed in reaction mixture containing superoxide and H₂S in DMSO.⁹⁷

Table 2.

Rate Constants for the Reaction of H₂S with Biologically-Relevant Oxidants

	reduction potential			kinetics of reaction with H2S
oxidant	couple	E°′ (V)	ref	k(M−1 s−1)	ref
One-Electron Oxidant
hydroxyl radical	HO•, H+/H2O	+2.31	414	1.1 × 1010 (pH 7)	415,416
oxygen	O2(g)/O2•−	−0.35a	414	very slow
carbonate radical	CO3•−, H+/HCO3−	+1.77b	63	2.0 × 108 (pH 7, 20 °C)	65
nitrogen dioxide	NO2•/NO2−	+1.04	63	1.2 × 107 (pH 7.5, 25 °C)	61
superoxide radical	O2•−, 2H+/H2O2	+0.91	414	~208 (DMSO)	97
myeloperoxydase compound I	CI/Fe3+	+1.35	421	1.1 × 106 (pH 7.4, 25 °C)	376
myeloperoxydase compound II	CII/Fe3+	+0.97	421	2.0 × 105 (pH 7.4, 25 °C)	376
Two-Electron Oxidant
hydrogen peroxide	H2O2, 2H+/2H2O	+1.35	414	0.48–0.73 (pH 7.4, 37 °C)	61,73
peroxynitrite	ONOOH, H+/NO2−, H2O	+1/30	422	6.7 × 103 (pH 7.4, 37 °C)	420
hypochlorite	HOCl, H+/Cl−, H2O	+1.28	421	0.8–20 × 108 (pH 7.4, 37 °C)	61,418
tauramine-chloramine	–			303 (pH 7.4, 37 °C)	61

^aThe reduction potential based on a standard state of 1 M O₂ is −0.18.⁴¹⁴

^bExtrapolated to pH 7 from E°(CO₃^•−/CO₃²⁻) = 1.57 V assuming a pK_a of 10.32 for HCO₃⁻.

The initial oxidation product of H₂S is the sulfyil radical (HS^•). HS^• is an oxidizing free radical capable of reacting with electron donors including ascorbate and GSH. Importantly, the one-electron oxidation of H₂S can unleash oxygen-dependent free radical chain reactions amplifying the initial oxidative event.^61,65 Although the reaction of HS^• with a second HS^• to form HSSH has a high rate constant ((6–9) × 10⁹ M⁻¹ s⁻¹, eq 7),⁶⁵ this reaction is unlikely to occur in most contexts because of its dependence on the square of HS^• concentration. Alternatively, HS^• can react with O₂ to form SO₂^•− ((5–7) × 10⁹ M⁻¹ s⁻¹, eq 10),⁶⁵ a reducing radical which in turn can react with O₂ forming O₂^•−. HS^• can also react reversibly with HS⁻ forming HSSH^•− (forward and reverse rate constants, 5.4 × 10⁹ M⁻¹ s⁻¹ and 5.3 × 10⁵ s⁻¹);⁶⁵ the latter can also react with O₂ forming O₂^•− (eq 11 and 12).^65,97,415 Superoxide radical (O₂^•−) can dismutate, spontaneously or enzymatically, to O₂ and H₂O₂⁴¹⁴

The rate constants for reaction of H₂S with two-electron oxidants (Table 2) are also comparable to those of low molecular weight thiols.⁶¹ The reaction with hydroperoxides (ROOH) initially forms HSOH, which can react with a second HS⁻ to form HSSH.^61,73 In the case of hydrogen peroxide, the final products depend on the initial ratio of hydrogen peroxide to H₂S and consist mainly of polysulfides, elemental sulfur, and, in the presence of excess oxidant, sulfate.^73,417 By analogy to thiols, the reaction with hypochlorous acid is likely to form HSCl that quickly hydrolyzes to HSOH.⁴¹⁸

The reaction of peroxynitrite with H₂S is more complex than its reaction with thiols and generates novel products.^61,419,420 The decay of peroxynitrite in the presence of H₂S is first order in peroxynitrite and first order in H₂S; the second order rate constant is 6.7 × 10³ M⁻¹ s⁻¹ (pH 7.4, 37 °C).⁴²⁰ The pH-dependence is bell-shaped, consistent with HS⁻ and ONOOH being the reacting species. Computational modeling suggests that the reaction starts with the nucleophilic substitution of HS⁻ on ONOOH to give HSOH and NO₂⁻ as initial products. The reaction then proceeds to the formation of “yellow” products that absorb at 408 nm.⁴¹⁹ The increase in absorbance at 408 nm occurs with a lag phase, consistent with the formation of intermediates that precede formation of the yellow products.⁴²⁰ Free radical scavengers or nitrite had no effect on the amount of yellow product formed, but the yield increased when peroxynitrite was in excess. Thus, it was proposed that the reaction of HSSH with peroxynitrite leads to formation of the yellow products, and indeed, mixtures of HSSH and peroxynitrite in acetonitrile yielded products with similar absorbance spectra.⁴²⁰ Based on mass spectrometric and computational studies, it was proposed that at least one of the yellow products is HSNO₂ or its isomer HSONO. In addition to the direct reaction of peroxynitrous acid with HS⁻, the free radicals derived from peroxynitrite (nitrogen dioxide, hydroxyl and carbonate radical) can also react with H₂S.⁶¹

The probability of H₂S acting as a direct scavenger of oxidants in biological systems depends on kinetic factors, i.e., on the products of the rate constants times H₂S concentration. While the reactions of H₂S with some oxidants display relatively high rate constants, comparable to those of LMW thiols, the tissue concentrations of H₂S (submicromolar, see section 3.5) are very low, i.e., several orders of magnitude lower than those of other reductants (e.g., millimolar for some thiols). Thus, it can be concluded that the direct reaction of H₂S with oxidants would not be fast enough in biological contexts to support a significant scavenging role. Furthermore, H₂S would not be able to compete with thiols for one- and two-electron oxidants, unless high local concentrations H₂S were reached as, for example, with bolus administration of exogenous H₂S. In conclusion the biological “antioxidant” effects ascribed to H₂S are unlikely to be due to direct scavenging of oxidants by H₂S but rather to indirect effects on enzymes, transporters, and/or other targets in signaling pathways.

6.3. Reaction of H₂S with NO^• and Its Metabolites

NO^• has important signaling roles in mammals including blood pressure regulation,^423,424 immune defense,^425,426 and neurotransmission.^427–429 Most of the “classical” effects of NO^•, such as vasodilation or neuro-modulation, are mediated by coordination of NO^• to the heme iron in sGC, which activates the enzyme to generate cyclic guanosine monophosphate (cGMP), a powerful second messenger.^430,431 But not all of the actions of NO^• proceed via cGMP signaling (Chart 20). NO^• can also lead to an oxidative posttranslational modification of cysteine called S-nitrosation.^432–436 How S-nitrosothiols are formed in the cells is still a matter of debate.^436–438 NO^• can also undergo one-electron reduction to form nitroxyl (HNO, IUPAC name: hydridooxidonitrogen, azanone, nitrosylhydride),^439–442 a powerful vasodilator.^267,439,442 NO^• is oxidized to nitrite and nitrate. Nitrite is now recognized as an important metabolite that can be reduced to NO^•.^443–445 Finally, peroxynitrite and its protonated form (ONOOH) can be generated in a diffusion controlled reaction between NO^• and O₂^•− (∼1 × 10¹⁰ M⁻¹ s⁻¹).^446–448

Chart 20. Interaction of H₂S with NO^• and Its Metabolites^a.

^aNO^• signals via the classical soluble guanylate cyclase (sGC)/cyclic GMP (cGMP) cascade. H₂S can reduce sGC to increase NO^• binding and cGMP production. cGMP is deactivated by phosphodiesterase 5 (PDE), an enzyme that is inhibited by H₂S. Oxidation of NO^• to nitrosonium (NO⁺) ion leads to modification of cysteine residues and formation of S-nitrosothiols (RSNO), a process that H₂S can facilitate. NO^• can be reduced by H₂S to form HNO. HNO activates the release of calcitonin gene-related peptide (CGRP), a vasodilator, but HNO can also be trapped by H₂S. NO^• is oxidized to nitrite, which can be reduced back to NO^•, a process that H₂S can facilitate. NO^• reacts with superoxide to form peroxynitrite (ONOO⁻). H₂S can scavenge ONOO⁻.

H₂S interferes with NO^• signaling, either by reacting with NO^• or its downstream metabolites^{267,419,420,449–453} or by modulating NO^• production^268,454,455 and cGMP levels.^266,377,410 The first report on H₂S-induced vasodilatory effects demonstrated its synergy with NO^•.⁷ Inhibition of endothelial NO synthase (eNOS) leads to abrogation of H₂S-induced vasodilation,^7,266,267 while deletion of CSE prevented the vasodilatory effects of acetylcholine and NO^•.²⁶⁶ In addition, the cardioprotective effects of H₂S were abolished in eNOS^−/− mice.²⁶⁸ Different mechanisms have been proposed for this crosstalk that are covered in section 10. In this section, we focus on the chemical aspects of the direct reactions between NO^• (and its metabolites) and H₂S.

6.3.1. Direct Reaction between NO^• and H₂S

Studies on the direct reaction between NO^• and H₂S date back to the 19th and early 20th century. The reaction of gaseous NO^• and H₂S was reported to form, among other products, nitrous oxide (N₂O) and elemental sulfur.^456–461 Formation of N₂O was difficult to explain as a single step process. N₂O is a product of HNO dimerization^439–442 so HNO formation could be the actual intermediate step.⁴⁶¹

It has been suggested that NO^• and H₂S can form HNO in vivo.^451,452 The combination of H₂S and NO^• donors was observed to have the same effects in murine heart as the application of the HNO donor, Angeli’s salt.⁴⁵² Indeed, when the reaction between H₂S and NO^• was studied at pH 7.4 under anaerobic conditions, the rate of HNO formation was first order on both NO^• and H₂S 267. The combination of NO^• and H₂S (2 μM each) yielded a peak HNO concentration of ∼0.5 μM, similar to the effects of 1 mM Angeli’s salt. Intracellular HNO production, detected by an HNO fluorescence sensor was also shown to depend on both NO^• and H₂S.²⁶⁷

Direct one-electron transfer from HS⁻ to NO^• to give HNO and S^•− (eq 22) is thermodynamically unfavorable (ΔG⁰′ = +102 kJ/mol).⁶⁴ An alternative mechanism is

$${\text{NO}}^{•}+{\text{HS}}^{-}\to \text{HNO}+{\mathrm{S}}^{•-}$$ (22)

the formation of HSNO^•− (eq 23), which is similar to the reaction

$${\text{NO}}^{•}+{\text{HS}}^{-}\to {\text{HSNO}}^{•-}$$ (23)

reported between NO^• and aromatic and “pseudoaromatic” alcohols such as tyrosine, hydroquinone, and ascorbic acid.⁴⁶² HSNO^•− is a powerful reducing agent (RSNO^•−/RSNO, E < −1 V)⁴⁶³ that can initiate a cascade of reactions, leading to N₂O and S_n formation (eqs 24–28).

$${\text{HS}}^{-}+{\text{NO}}^{•}\to {\text{HSNO}}^{•-}$$ (24)

$${\text{HSNO}}^{•-}+{\text{NO}}^{•}+{\mathrm{H}}^{+}\to \text{HSNO}+\text{HNO}$$ (25)

$$\text{HSNO}+{\text{HS}}^{-}\to \text{HNO}+{{\text{HS}}_2}^{-}$$ (26)

$${{\text{HS}}_2}^{-}+{\mathrm{H}}^{+}\to {\mathrm{H}}_2\mathrm{S}+\frac{1}{n}{s}_n$$ (27)

$$2\text{HNO}\to {\mathrm{N}}_2\mathrm{O}+{\mathrm{H}}_2\mathrm{O}$$ (28)

Endogenous HNO production would be critically dependent on NO^• and H₂S being produced in close proximity since both can engage in competing reactions with other molecules (Figure 15A). The transient receptor potential channel A1 (TRPA1), a biological sensor that regulates HNO-induced release of the powerful vasodilator calcitonin gene related peptide, colocalizes with CBS²⁶⁷ and nNOS,⁴⁶⁴ potentially forming a functional unit for HNO formation and its action.^267,465 Studies with an HNO-responsive two-photon ratiometric fluorescence imaging probe confirmed that endogenous HNO generation is dependent on endogenous H₂S and NO^• formation in cells and brain tissues.⁴⁶⁶

Figure 15.

Signaling aspects of NO^•/H₂S cross-talk. (A) To form HNO and minimize side reactions, H₂S and NO₂^• have to be produced in proximity. HNO (and possibly HSNO) reacts with protein thiols and glutathione. (B) All three gases, NO^•, O₂, and H₂S tend to accumulate in membranes. NO^• and O₂ form N₂O₃, which readily reacts with H₂S to form HSNO. (C) HSNO formed in the reaction of protein S-nitrosothiols with H₂S can diffuse through the cell membrane and transfer the “NO⁺” group to another protein target.

6.3.2. Reaction of H₂S with S-Nitrosothiols and Metal-Nitrosyls

Formation of a new S-nitrosothiol, HSNO (IUPAC name: nitrososulfane or (hydridosulfanido)oxidonitrogen), in the reaction of H₂S with S-nitrosothiols (or other nitrosocontaining species) was first proposed in 2006.^467,468 HSNO was known as a product of cis-HNSO photolysis in argon matrices, where it has been studied computationally and by IR spectroscopy.^469–472 The crystal structure of the bis-(triphenylphosphine)iminium SNO salt (PNP⁺SNO⁻) was reported,⁴⁷³ but a detailed study in aqueous solution was missing.

Using pulse radiolysis to generate HS^• and NO^•, formation of a species with the spectral characteristics (λ_max ≈ 330 nm) of an S-nitrosothiol was observed (Chart 21).⁴⁴⁹ The species was short-lived with a half-life of ∼12 μs⁴⁴⁹ yielding an estimate of ∼10⁷ M⁻¹ s⁻¹ for the rate constant for the reaction of HSNO with sulfide anion (eq 26).⁶⁴

Chart 21.

Reaction Paths for HSNO Generation (1–3) and Its Biologically Relevant Reactions (4–6)

HSNO was also detected in reactions of H₂S with “NO⁺” carriers: acidified nitrite,⁴⁴⁹ N₂O₃,⁴⁷⁴ metal nitrosyls,^{450,475–477} and S-nitrosothiols.^449,453 For example, in the reaction with acidified nitrite, a brown-red intermediate was formed prior to the solution turning milky white. ESI-TOF MS analysis of the acidic and neutralized solution of the brown-red intermediate revealed the parent ion mass and isotopic pattern expected for HSNO.⁴⁴⁹

Reaction of thiosemicarbazide with NO^• results in HSNO formation under physiological conditions; thiosemicarbazides are therefore proposed to serve as a tool capable of transforming intracellular NO^• into HSNO.⁴⁷⁸

In the reaction of H₂S with N₂O₃ (eq 29), the facile formation of stable HSNO was demonstrated by Fourier-transform microwave spectroscopy.⁴⁷⁴

$${\mathrm{N}}_2{\mathrm{O}}_3+{\mathrm{H}}_2\mathrm{S}\to \text{HSNO}+{\text{HNO}}_2$$ (29)

Generation of HNO and N₂O (eq 26 and 28) was confirmed when ¹⁸O labeled N₂O₃ was reacted with excess of H₂S.⁴⁷⁴ HSNO formation from N₂O₃ and H₂S could be important for intracellular RSNO generation. Formation of N₂O₃ is deemed to be kinetically improbable due to the low intracellular concentration of NO• compared to O₂.⁴³⁷ However, N₂O₃ could be formed in the lipid bilayers where NO^• and O₂ accumulate.⁴³⁷ H₂S can also accumulate in lipid bilayers based on its partition coefficient⁵⁷ creating conditions that might be conducive for HSNO formation (Figure 15B). Reaction of HSNO with thiols (Chart 21 and Figure 15C) can result in transnitrosation. HSNO can act as an “NO⁺” carrier from one protein to another and across the cell membrane (Figure 15C). This idea is supported by the ability of H₂S to promote nitrosation of BSA outside a dialysis bag containing nitrosated BSA but also nitrosation of hemoglobin in the red blood cells from extracellular nitrosated BSA.⁴⁴⁹ NaHS treatment during cardiac ischemia was reported to increase tissue S-nitrosation⁴⁷⁹ possibly via HSNO formation which would then act as transnitrosating agent.

Using a combination of spectroscopic approaches and ESI-TOF MS to study the trans-nitrosation reaction between S-nitrosoglutathione and H₂S, HSNO was detected within 30 min.⁴⁴⁹ Greater than equimolar H₂S concentration promoted N₂O and hydroxylamine formation via intermediate HNO generation (eqs 26, 28, and 30).

$$\text{HNO}+2{\text{HS}}^{-}+{\mathrm{H}}^{+}\to {\text{NH}}_2\text{OH}+{{\text{HS}}_2}^{-}$$ (30)

An unexplained observation made during the transnitrosation reaction between GSNO and H₂S was that the solution turned yellow (λ_max = 412 nm).^449,480 MS analysis of the RSNO/H₂S reaction mixture identified, among other products, SSNO⁻ which was suggested to be a stable yellow product.⁴⁸¹ SSNO⁻ formation from RSNO and H₂S was proposed to occur via eqs 31–33.^453,480,481

$$\text{RSNO}+{\text{HS}}^{-}\to {\text{SNO}}^{-}+\text{RSH}$$ (31)

$${\text{SNO}}^{-}+{\text{HS}}^{-}\to {{\text{HS}}_2}^{-}+\text{HNO}$$ (32)

$${{\text{HS}}_2}^{-}+“{\text{NO}}^{+}”(\text{RSNO}\text{or}\text{HSNO})\rightleftharpoons {\text{SSNO}}^{-}+\text{RSH}(\text{or}{\mathrm{H}}_2\mathrm{S})$$ (33)

Whether SSNO⁻ is stable enough to mediate biological effects has been the subject of debate. Crystalline PNP⁺SSNO⁻ synthesized by a published method⁴⁷³ has been used to chemically characterize SSNO⁻.⁶⁹ Crystalline SSNO⁻ and SSNO⁻ dissolved in organic solvent are air and water sensitive.^69,473 The reduction potential of SSNO⁻ was determined to be −0.21 V versus NHE, which is within the range of physiological reductants like glutathione with which it reacted readily.⁶⁹ Furthermore, SSNO⁻ decomposed rapidly in the presence of H₂S and cyanide forming SNO⁻.^{69,482 15}N/¹⁴N NMR and cryo-ESI TOF MS analyses confirmed the formation of SNO⁻/HSNO in the reaction of SSNO⁻ with HS⁻ and CN⁻ (eqs 34 and 35).⁴⁸²

$${\text{SS}}^{15}{\text{NO}}^{-}+{\text{HS}}^{-}\to {\mathrm{S}}^{15}{\text{NO}}^{-}+{{\text{HS}}_2}^{-}$$ (34)

$${\text{SS}}^{15}{\text{NO}}^{-}+{\mathrm{C}}^{14}{\mathrm{N}}^{-}\to {\text{SC}}^{14}{\mathrm{N}}^{-}+{\mathrm{S}}^{15}{\text{NO}}^{-}$$ (35)

A role for SSNO⁻ in signaling is doubtful based on kinetic grounds (the apparent rate constant for its formation estimated from published data⁴⁸¹ is 10⁻¹⁴ M⁻¹ s⁻¹) as well.⁶⁴ Considering the very low concentrations of HS⁻ (section 3.5) and RSNO^436,437 and the very high concentrations of thiol, the reaction of SNO⁻ with HS⁻ (eq 32) in a cellular millieu seems unlikely. Furthermore, HS₂⁻, which is intrinsically unstable and readily reduced by thiols, is unlikely to persist long enough or to react specifically with “NO⁺” to form SSNO⁻ (eq 33). Even if formed, SSNO⁻ would readily react with thiols to form HSNO.⁴⁸² In summary, HSNO remains the chemically most plausible nitrosating agent that can react with cysteines (Chart 21 and Figure 15) and engage in transnitrosation reactions.^482,483

6.3.3. Metal-Catalyzed Reaction between Nitrite and Sulfide

Although nitrite and H₂S do not react directly at pH 7.4,⁹⁷ NO^• generation has been reported in cells treated with these two reagents.⁴⁵⁰ Intracellular HNO generation which was localized to mitochondria was observed when cells were treated with 100 μM nitrite and sulfide but not in cells depleted of mitochondria.⁴⁵⁰ This result implicated a role for mitochondrial proteins in catalyzing the reaction between nitrite and H₂S. To understand the possible role of heme iron in the reaction mechanism for HNO generation from nitrite and sulfide, a water-soluble iron–porphyrin was used as a model system.⁴⁵⁰ Initial binding of nitrite to ferric heme and subsequent oxygen atom transfer^484,485 to H₂S to give HSOH was the predominant reaction observed when nitrite was in excess. The [Fe^II(NO)] ↔ [Fe^III(NO⁻)] complex then slowly released HNO.^450,485 When sulfide was in excess, it reduced ferric to ferrous heme so that the classic nitrite reductase activity of Fe^II heme was observed. The formed [Fe^III(NO)]↔[Fe^II(NO⁺)] reacted with HS⁻ to form an [Fe^II(HSNO)] complex (Chart 22).⁴⁵⁰ These results support the feasibility of H₂S reacting with metal-nitrosyls to form S-nitrosothiols via HSNO, which represents an alternative mechanism for the physiological generation of HNO.

Chart 22. Proposed Reaction Mechanism for H₂S-Assisted Nitrite Reduction Catalyzed by an Iron Porphyrin Compound^a.

^aPathway A, which predominates when nitrite is in excess over H₂S, represents a classical oxygen atom transfer; nitrite coordination to ferric heme leads to HSOH and [Fe²⁺(NO)] ↔ [Fe³⁺(NO⁻)], which releases HNO slowly. Pathway B predominates when H₂S is in excess over nitrite; reduction of ferric heme by H₂S is followed by nitrite reduction to [Fe³⁺(NO)] ↔ [Fe²⁺(NO⁺)] species, which is scavenged by HS⁻ giving HSNO. Either free or coordinated HSNO causes transnitrosation of protein thiols or, in the reaction with H₂S, generates HNO.

Another metalloprotein, a molibdopterin-containing xanthine oxidase, was reported to catalyze H₂S-stimulated nitrite reduction in endothelial cells and in mice injected with Na₂S. However, the mechanism of this reaction was not elucidated.⁴⁸⁶

7. PROTEIN PERSULFIDATION

Protein persulfidation, an oxidative posttranslation modification of cysteines, represents a mechanism by which H₂S signals. This modification is also referred to in the literature as “sulfhydration”,¹⁹ which implies “hydration” and is inaccurate. Instead, the process involves “sulfuration”, i.e., the addition of a sulfur atom.^64,487 The term “persulfuration” has been also used, but the term “persulfidation” has been gaining wide acceptance and is used here. Other ways to describe RSSH are hydropersulfide, or hydrodisulfide, or as a disulfane derivative (e.g., CH₃SSH is methyldisulfane⁶⁴). A less ambiguous name for RSSH is hydridodisulfide. In this review, the term “persulfide” is used to designate RSSH/RSS⁻.

Contrary to the chemically incorrect claim that H₂S can directly modify cysteine residues to form persulfides, the reaction between H₂S and thiols (eq 36) requires an oxidant.^64,181

$$\text{RSH}+{\text{HS}}^{-}\to {\text{RSS}}^{-}+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}{E}^{\circ \prime }=+0.18\mathrm{V}$$ (36)

Due to their instability and greater reactivity than thiols, working with persulfides is challenging. In the following subsections, progress on developing methods to study persulfides and our current understanding of their reactivity are discussed.

7.1. Model Systems to Study Protein Persulfidation

Persulfides are relatively unstable in aqueous solution and are typically synthesized immediately before use. Several model systems have been used to study persulfide chemistry. These models are grouped in two categories: (i) low molecular weight (LMW) persulfides and (ii) protein persulfide models.

7.1.1. LMW Persulfide Models

Several LMW persulfides have been reported that are either synthesized in situ or have been characterized following purification. A synthetic method for persulfides dates back to 1954 when alkyl- and arylpersulfides were prepared from sulfenyl chloride and thiols.⁴⁸⁸ The resulting acyldisulfide was hydrolyzed by HCl to give persulfide (Chart 23A, [I]). Persulfides can also be prepared from methoxycarbonyl disulfides, which would undergo alkoxide-induced displacement of the RSS⁻ anion (Chart 23A, [II]).⁴⁸⁹ Alternatively acyl disulfides can be synthesized in the reaction of dialkyl thiosulfones with thioacid (Chart 23A, [III]).⁴⁹⁰ In fact, the acidic hydrolysis of acyl disulfides has become a general synthetic strategy for the preparation of small molecule persulfides like ethyl-, t-butyl-, benzyl-, diphenylmethyl-, trityl-, adamantyl-, and penicillamine-derived persulfide (Chart 23B).^{488,491–500} The hydrophobic LMW persulfides need to be handled in organic solvents and are consequently protonated,^494,495 which reduces their nucleophilicity. In contrast, persulfides are deprotonated and more reactive at physiological pH (see section 6.2).

Chart 23.

Synthetic Strategies for the Preparation of Low Molecular Weight Persulfides (A) and Structures of Some of the Persulfides Prepared following These Synthetic Routes (B)

A water-soluble penicillamine-derived LMW persulfide has been prepared.⁵⁰¹ The synthetic protocol involved acylprotected disulfides and gave high yields (Chart 24). At pH 2.7, <5% degradation of the acyl-protected disulfide of penicillamine was observed after 120 min at room temperature, allowing relatively stable stock solutions to be prepared. When placed in buffers with pH >6 the disulfide underwent S- to N-methoxycarbonyl transfer, generating N-methoxycarbonyl penicillamine persulfide, a persulfide related to a commonly used S-nitrosothiol (Chart 24).

Chart 24.

Synthesis of an Acyl-Protected Disulfide of Penicillamine (A) and Its Rearrangement to N-Methoxycarbonyl Penicillamine Persulfide via S- to N-Methoxycarbonyl Transfer at pH 7.4 (B)

LMW persulfides can be generated in situ in aqueous solution by mixing disulfides (such as cystine or GSSG) with H₂S in equimolar ratio (Chart 25A).^60,502–505 As this is an equilibrium process, the reaction mixture contains unreacted disulfides and H₂S in addition to the persulfide. Alternatively, persulfides can be prepared in situ by CBS- or CSE-catalyzed conversion of cystine or homocystine to the corresponding persulfides.^165,226 Cysteine persulfide and homocysteine persulfide are formed via α,β or α,γ elimination reactions, respectively (Chart 25B; for details see section 3). The LMW persulfide, GSSH, can also be formed in situ. Rhodanese in the presence of thiosulfate^313,330 (or p-toluenethiosulfonate)^506,507 and glutathione forms GSSH (Chart 25C) while SQR forms GSSH via sulfurtransfer from H₂S to GSH (Figure 8).³¹¹ Another route for preparing GSSH is to reduce the trisulfide (GSSSG) with glutathione reductase and NADPH (Chart 25D).^165,508

Chart 25. Strategies for the in Situ Preparation of LMW Persulfides^a.

^a(A) Cysteine and glutathione persulfides can be prepared by the reaction of H₂S with cystine and glutathione disulfide, respectively. (B) Cysteine persulfide can be generated from cystine and CSE or CBS. (C) Rhodanese can be used to transfer sulfur to glutathione. (D) Glutathione reductase (GR) uses electrons from NADPH to reduce glutathione trisulfide to persulfide and GSH.

7.1.2. Protein Persulfide Models

A commonly used strategy for preparing protein persulfides is the reaction of activated disulfides with equimolar H₂S. For example, a protein with one reactive cysteine is first treated with Ellman’s reagent, 5,5′-dithiobis(2-nitrobenzoate) (DTNB), to form a mixed disulfide.^60,505,509 The thionitrobenzoate anion (TNB) is a good leaving group, and in the next step the protein–TNB mixed disulfide is reacted with an equimolar concentration of H₂S to generate the protein persulfide (Figure 16A).^60,509 The concomitant release of TNB, which has a strong absorbance at 412 nm,⁵¹⁰ provides a simple method for quantifying the reaction yield. Persulfides of papain,^505,509 glutathione peroxidase 3,⁵⁰⁹ and human serum albumin⁶⁰ have been prepared using this approach.

Figure 16.

Strategies for the preparation of protein persulfides. (A) Protein thiols react first with DTNB forming mixed disulfides that react with H₂S forming persulfides. Thionitrobenzoate is a good leaving group and its UV–visible absorbance can be used to estimate the yield of persulfides. (B) Protein sulfenic acids, when stable, can be used as precursors for persulfide preparation. (C) Protein thiols can be mixed with inorganic polysulfides or with a mixture of HOCl and H₂S. Besides persulfides, polythiolated products are also formed. (D) Protein thiols can react with 9-fluorenylmethyl disulfide. The products undergo alkaline hydrolysis forming persulfides.

The reaction of sulfenic acid with H₂S (for details see section 8.2) can also be used to prepare protein persulfides.⁵¹¹ The primary obstacle with this approach is that sulfenic acid modifications on proteins are generally unstable. An exception is the sulfenic acid derivative of serum albumin, which is relatively stable^512,513 and has been exploited to generate the corresponding persulfide (Figure 16B).^60,511

A less specific approach for protein persulfidation involves mixing the protein with H₂S and an oxidant, such as HOCl, or mixing the protein with polysulfide salts (Figure 16C).⁵¹⁴ Uncontrolled protein poly thiolation is an inevitable outcome of this approach (see section 8.3), and the use of these methods is discouraged.

Alternatively, a protein thiolate can be reacted with 9-fluorenylmethyl disulfide to form a mixed disulfide which is then exposed to alkaline pH to promote hydrolysis generating the protein persulfide (Figure 16D).⁵¹⁵ However, the alkaline conditions could lead to protein denaturation.

7.2. Persulfide Reactivity

Persulfides have characteristics in common with thiols, disulfides, polysulfides, hydroperoxides, and sulfenic acids. The chemistry of persulfides is very rich, and persulfides are very versatile molecules that are being assigned roles of increasing importance in biology.

The crystal structure of tritylpersulfide shows an S–S bond length of 2.0396 Å, which fits well with the S–S bond length observed in crystals of inorganic polysulfides. The CSSH dihedral angle is 82.2°,⁴⁹⁴ which is close to the CSSC dihedral angle of 83° seen in unstrained disulfides.⁵¹⁶ The crystal structures of some proteins involved in sulfur metabolism have been obtained with cysteine residues modified to persulfides, e.g., the structures of bovine rhodanese^332,517 and of human MST.²⁷¹ In addition, the crystal structures of some proteins that contain free cysteine persulfide bound as a ligand have been obtained.^518,519

In alkaline solutions, persulfides show an absorption maximum at 335–340 nm^{494,495,501,502} and a relatively low absorption coefficient (∼310 M⁻¹ cm⁻¹).³¹ The IR spectra of alkyl and aryl persulfides show a weak S–H stretch at ∼2500 cm⁻¹ (Table 3), which is shifted to lower wavenumbers than thiols (∼2570 cm⁻¹), consistent with the presence of a stronger S–H bond in thiols.^494,495,520 Similar shifts are observed in the Raman spectra, with the additional presence of a band in the 200–500 cm⁻¹ region due to the S–S bond. ¹H NMR spectra of persulfides in organic solvents show shifts in the S–H proton relative to the corresponding thiols. For example, the S–H proton shows an ∼0.4 ppm upfield shift in tritylpersulfide and an ∼1.2 ppm downfield shift in benzenepersulfide and adamantylpersulfide (Table 3).^494,495

Table 3.

Basic Physicochemical and Thermodynamic Properties of LMW Persulfides

	values	refs
λmax	335–340 nma	31,494,495,501,502
IR (S–H stretch) (cm−1)	2490–2510	488–500
(S–S stretch) (cm−1)	200–500
1H NMR	2.7–3 ppm	494,495,501,502
S–S bond length	2.04 Å	495
E°′(RSS−, 2H+/RSH, HS−)	−0.18 Vb	64
E°′(RSS•/RSS−)	+0.68 Vb	64

^aAlkaline pH, but organic solvents as well.

^bVersus SHE.

7.2.1. Persulfide Acidity

Protonated persulfides can ionize to form the corresponding anionic persulfides (eq 37).

$$\text{RSSH}\rightleftharpoons {\text{RSS}}^{-}+{\mathrm{H}}^{+}$$ (37)

In agreement with the weaker S–H bond in persulfides than in thiols, the acidity of persulfides is predicted to be higher. There are very few reported experimental measurements of persulfide pK_a. From the pH dependence of the rate of hydrogen atom transfer from 2-[(3-aminopropyl)amino]ethane persulfide to a carbon-centered free radical, the pK_a of the persulfide was estimtated to be 6.2 ± 0.1 in comparison to a pK_a of 7.6 ± 0.1 for the corresponding thiol.⁴⁸⁹ A computational study estimated that the pK_a of cysteine persulfide (4.3) is ∼4 units lower than of cysteine thiol (8.29).⁶⁰ The available data suggests that at physiological pH, RSS⁻ will predominate over RSSH and that the [RSS⁻]/[RSSH] ratio can be ∼10⁴-fold higher than the corresponding [RS⁻]/[RSH]. The acidity of persulfides and thiols on proteins will be modulated by their microenvironment, i.e., by the presence of functional groups.

7.2.2. Persulfide Nucleophilicity

Ionized persulfides (RSS⁻) are nucleophilic. Although both sulfurs in the ionized and in the protonated species have lone pairs of electrons, the outer or terminal sulfur in RSS⁻ is the more nucleophilic center (eq 38).

$${\text{RSS}}^{-}+{\mathrm{E}}^{+}\to \text{RSSE}$$ (38)

Basicity and nucleophilicity are generally correlated, and the stronger the base, the greater the nucleophilicity.⁵²¹ Since the pK_a of persulfides is significantly lower than that of the corresponding thiol,⁶⁰ persulfide would be expected to be less nucleophilic. However, the presence of a vicinal sulfur atom with lone electron pairs increases the nucleophilicity of the terminal sulfur atom via the alpha effect.^522,523 Examples of the alpha effect enhancing nucleophilicity include HOO⁻ relative to HO⁻ and NH₂NH₂ and NH₂OH relative to NH₃.⁵²¹ As noted previously, nucleophilicity is a kinetic concept and needs to be evaluated from the rate constants for the relevant reactions. A comparison between the reactivity of the persulfide versus thiol in human serum albumin toward 4,4′-dithiodipyridine provided a quantitative estimate of the magnitude of the alpha effect.⁶⁰ The pH-independent rate constant for the reaction of the albumin persulfide with 4,4′-dithiodipyridine was 3-fold greater than for the thiolate. At pH 7.4, the persulfide is estimated to be fully ionized while the thiol is only partially ionized (the pK_a of the single thiol is 8.1).⁶⁰ Thus, the observed rate constant at pH 7.4 was 20-fold greater for persulfide than for the thiol.⁶⁰

Computational evaluations of the HOMO energies are consistent with a higher nucleophilicity of persulfides than thiols. In one estimate, the energy of the HOMO of methylpersulfide was ∼29 kJ mol⁻¹ higher than that of methylthiolate.¹⁰³ In another estimate, the HOMO of cysteine persulfide was 51 kJ mol⁻¹ higher than that of cysteine thiolate and, in addition, cysteine persulfide had lower chemical hardness than cysteine thiolate.⁶⁰

The nucleophilicity of persulfides is evident from their reactivity toward thiol alkylating agents such as 1-chloro-2,4-dinitrobenzene,⁴⁹⁹ iodoacetamide,^505,509,515 N-ethylmaleimide,^505,509,515 methyl acrylate,⁵¹⁵ monobromobimane,^165,515 benzyl bromide,⁵¹⁵ 2-methylsulfonyl benzothiazol,^{135,511,515,524} and methylmethanethiosulfonate⁵⁰⁹ (Chart 26A). In contrast to thiols, which form thioeters with these reagents, persulfides form disulfides, which can be reduced to the corresponding thiols with reductants such as dithiothreitol. Interestingly, 9-fluorenylmethyl-based thioester models of LMW persulfide reacted with methylsulfonyl benzothiazol to give a trisulfide due to the reactivity of the initial benzothiazol disulfide derivative,⁵¹⁵ which was not the case for other LMW persulfides or for protein (bovine serum albumin and glutathione peroxidase-3) persulfides.^135,511,524 The nucleophilicity of persulfides is also revealed by their reaction with disulfides such as 5,5′-dithiobis(2-nitrobenzoic acid),^60,509 N-acetylcysteine piridyldisulfide,⁵⁰⁹ and 4,4′-dithiodipyridine,⁶⁰ to form trisulfides and other products (Chart 26B).

Chart 26. Reactions of Persulfides with Electrophiles^a.

^a(A) Reactions with thiol alkylating agents. (B) Reactions with disulfides. (C) Reactions with methylmercury and nitroguanosine 3′,5′-cyclic monophosphate.

Persulfides also react with electrophiles such as 8-nitroguanosine 3′,5′-cyclic monophosphate (8-nitro-cGMP) giving HS-cGMP,^165,501 with methylmercury⁵²⁵ (Chart 26C) and, as described in the following sections, with one- and two-electron oxidants. In summary, persulfides are better nucleophiles than thiols because of the greater availability of RSS⁻ versus RS⁻ at neutral pH and higher intrinsic reactivity due to the alpha effect.

7.2.3. Reaction of Persulfides with Two-Electron Oxidants

Another manifestation of the nucleophilicity of the persulfides is their reactivity with two-electron oxidants. For example, the apparent rate constant of the reaction of albumin persulfide with peroxynitrite (1.2 × 10⁴ M⁻¹ s⁻¹ at 20 °C) is 4-fold higher than of the corresponding thiol (2.7 × 10³ M⁻¹ s⁻¹).⁶⁰ By analogy with thiols, the immediate product of the reaction between a persulfide and a hydroperoxide is likely to be an unstable perthiosulfenic acid (RSSOH), which undergoes further reactions forming polysulfides (RS_nR and RS_n⁻) and, in the presence of excess oxidant, perthiosulfinic and perthiosulfonic acids (RSSO₂H and RSSO₃H). The latter have been detected as oxidation products of papain, albumin, and glutathione peroxidase.^60,509,511 Importantly, in contrast to the thiol-derived sulfinic and sulfonic acids (RSO₂H and RSO₃H), which are generally considered to be irreversible oxidation products, the corresponding persulfide derivatives RSSO₂H and RSSO₃H can be reduced back to thiol by common reductants.^135,340 Higher recovery of thiols after exposure of persulfides versus thiols to hydrogen peroxide (or HNO) followed by dithiothreitol treatment was demonstrated.⁵²⁶

The higher reactivity of persulfides to oxidants and the recovery of thiols following reduction of the resulting oxidation products, support the proposal that persulfides can serve a protective functions for protein thiols (Chart 27).^{135,340,341,527}

Chart 27. Protein Persulfidation Can Protect Proteins from Overoxidation^a.

^aA thiol can be oxidized to a sulfenic acid. The latter can be reduced back to thiol or be further oxidized to sulfonate, an irreversible modification. Persulfides, if exposed further to oxidants, will form S-sulfocysteines (−SSO₃⁻). Enzymes such as thioredoxin can reduce the S–S bonds and restore the native thiol.

7.2.4. Reaction or Persulfides with One-Electron Oxidants

Persulfides are excellent one-electron reductants and are in fact better than thiols or H₂S.⁵²⁸ This is consistent with the lower energy of dissociation of the S–H bond (293 kJ mol⁻¹) in comparison to thiols and H₂S (385 kJ mol⁻¹).⁵²⁹ It is also consistent with the one-electron reduction potential of persulfides (E°′(RSS^•/RSS⁻) = 0.68 V), which is lower than those of the corresponding thiol, (E°′(RS^•, H⁺/RSH) = 0.96 V) and of H₂S (E°′(S^•−, H⁺/HS⁻) = 0.91 V),⁶⁴ respectively. Thus, persulfides can be oxidized by weaker oxidants than thiols or H₂S, and RSS^• is less oxidizing than RS^• or HS^•.

Depending on the nature of the one-electron oxidant (A₁^• or A₂^• in eqs 39 and 40, e.g., carbon centered radical or peroxyl CCl₃OO^• radicals, respectively) the persulfide can react through a hydrogen atom or electron transfer mechanism, which affects the pH dependence of the process.⁵²⁸

$$\text{RSSH}+{{\mathrm{A}}_1}^{•}\to {\text{RSS}}^{•}+{\mathrm{A}}_1\mathrm{H}$$ (39)

$${\text{RSS}}^{-}+{{\mathrm{A}}_2}^{•}\to {\text{RSS}}^{•}+{{\mathrm{A}}_2}^{-}$$ (40)

Exposure of aralkyl persulfides to ferric salts under organic solvents led to the formation of tetrasulfide and ferrous ion, showing that persulfides can be oxidized by ferric ions (eqs 41 and 42).⁵³⁰

$$\text{RSSH}+{\text{Fe}}^{3+}\to {\text{RSS}}^{•}+{\mathrm{H}}^{+}+{\text{Fe}}^{2+}$$ (41)

$$2{\text{RSS}}^{•}\to \text{RSSSSR}$$ (42)

Importantly, persulfides were unable to oxidize ferrous salts.⁵³⁰ This contrasts with the behavior of hydrogen peroxide and alkyl hydroperoxides, which oxidize ferrous to ferric ion with concomitant formation of a hydroxyl radical via the Fenton reaction (eq 43).

$${\mathrm{H}}_2{\mathrm{O}}_2+{\text{Fe}}^{2+}\to {\text{HO}}^{-}+{\text{HO}}^{•}+{\text{Fe}}^{3+}$$ (43)

Persulfides can reduce ferricyanide,⁵³¹ ferric cytochrome c,^501,505,532 and metmyoglobin.⁵³¹ In addition, persulfides have been proposed to react with carbon centered radicals,⁴⁸⁹ peroxyl radicals,^528,533 and the nitroxide TEMPOL,⁵³¹ yielding the perthiyl radical. All of these reactions appear to be faster with persulfides than with thiols. For example, the rate constant for the reaction of 2-[(3-aminopropyl)amino]ethane persulfide and the carbon-centered α-hydroxyalkyl radical derived from isopropanol is 2.4 × 10⁹ M⁻¹ s⁻¹, which is 1 order of magnitude higher than the corresponding reaction of the thiol.⁴⁸⁹

Penicillamine-derived persulfide reduces ferric cytochrome c quantitatively in contrast to penicillamine and glutathione, which do not show significant reduction.⁵⁰¹ However, this reaction is thermodynamically uphill given the mismatch in redox potentials between the persulfide (+0.68 V) and cytochrome c (E°′(cyt c³⁺/cyt c²⁺) = +0.26 V).⁶⁴ Persulfides react directly with oxygen, albeit slowly (k < 0.3–0.4 M⁻¹ s⁻¹).¹⁰³ This reaction also faces thermodynamic and kinetic barriers due to a mismatch in the redox potentials and the triplet state of oxygen and reveals the intrinsically higher reactivity of persulfides with respect to thiols and H₂S. A likely explanation for why each of these unfavorable reactions occurs is that the perthiyl radical product is efficiently removed via recombination forming tetrasulfide (eq 44). Perthiyl radicals decay predominantly through second order processes with rate constants of 1–6 × 10⁹ M⁻¹ s⁻¹.^528,533,534 The resulting RSSSSR can decompose to give RSSR and S₂ (eq 45).

$$2{\text{RSS}}^{•}\to \text{RSSSSR}$$ (44)

$$\text{RSSSSR}\to \text{RSSR}+{\mathrm{S}}_2$$ (45)

Perthiyl radicals have been characterized by pulse radiolysis, flash photolysis, and EPR spectroscopy.⁵²⁸ The unpaired electron in perthiyl radicals is delocalized between the two sulfur atoms. The resonance stabilization energy of perthiyl radicals relative to thiyl radicals is estimated in 8.8 kJ mol⁻¹.^489,528 This inherent stability of perthiyl radicals contributes to the efficiency of persulfides as reductants.

In addition to the radical recombination reaction to form tetrasulfides, perthiyl radicals can also act as oxidants, but the rate constants of these reactions are smaller than those of thiyl radicals.⁵²⁸ For example, the rate constant for hydrogen atom abstraction by the perthiyl radical from a polyunsaturated fatty acid is 1.2 × 10⁶ M⁻¹ s⁻¹, 1 order of magnitude lower than the corresponding reaction of a thiyl radical, 1.4 × 10⁷ M⁻¹ s⁻¹. The rate constant for the reaction of a perthiyl radical with ascorbate is (1–6) × 10⁶ M⁻¹ s⁻¹, which is 1 order of magnitude smaller than that of the thiyl radical.⁵²⁸

Perthiyl radicals reportedly react with oxygen with a second order rate constant of 5 × 10⁶ M⁻¹ s⁻¹ initially forming an RSSOO^• species and ultimately forming inorganic sulfate.⁵³⁴ However, recent studies have not confirmed this experimentally, and computational studies suggest that the reaction of perthiyl radicals with oxygen is thermodynamically uphill.^531,533 In contrast, thiyl radical reacts reversibly with oxygen forming thioperoxyl radicals (RSOO^•) with forward and reverse rate constants of ∼10⁹ M⁻¹ s⁻¹ and ∼10⁵ s⁻¹, respectively,^535,536 a reaction that contributes to oxidative damage via chain propagation.

A perthiyl radical can react with RSS⁻ to form RSSSSR^•−, which is unstable,⁵³⁷ has not been detected directly, and reacts with oxygen to form O₂^•− and the more stable RSSSSR (eqs 46 and 47). Analogously, RSS^• can react with RS⁻ to form RSSSR^•−.⁵²⁸

$${\text{RSS}}^{-}+{\text{RSS}}^{•}\to {\text{RSSSSR}}^{•-}$$ (46)

$${\text{RSSSSR}}^{•-}+{\mathrm{O}}_2\to \text{RSSSSR}+{{\mathrm{O}}_2}^{•-}$$ (47)

In contrast to thiyl radicals that react with NO^• very rapidly forming nitrosothiols,⁵³⁸ perthiyl radicals do not appear to react with NO^•. The apparent lack of reactivity has been attributed to the relatively high stability of the radicals and to the weakness of the N–S bond in RSSNO. Accordingly, attempts to prepare RSSNO failed with immediate NO^• generation from the reaction mixture.^531,539

In summary, persulfides are excellent one-electron reductants, which can be explained by their ease of oxidation, by the high relative stability of the perthiyl radical, and by the rapid radical recombination, which facilitates product removal.

7.2.5. Electrophilicity

Persulfides are relatively weak electrophiles. The reactions of persulfides in the protonated state with a general nucleophile Nu⁻, are shown in eqs 48 and 49. When the inner sulfur is the site of nucleophilic attack, H₂S is released. When the outer sulfur is the site of attack sulfur transfer to the nucleophile occurs with elimination of the thiol.

$$\text{RSSH}+{\text{Nu}}^{-}\to \text{RSNu}+{\text{HS}}^{-}$$ (48)

$$\text{RSSH}+{\text{Nu}}^{-}\to \text{RSH}+{\text{NuS}}^{-}$$ (49)

Persulfides can react with cyanide,⁵⁴⁰ thiolates,^501,509,540 sulfite,⁵⁴⁰ phosphines,⁴⁹⁶ and amines.⁵⁴¹ Thiols and H₂S are formed as reaction products, together with S_n and polysulfides. In organic solvents, cyanide, amines, hydroxide, and halides react as bases rather than as nucleophiles, abstracting a proton from RSSH and promoting its decay.^{103,495,541,542}

Computational modeling of the lowest unoccupied molecular orbital (LUMO) of methylpersulfide (CH₃SSH) shows that attack at either sulfur atom is possible, while the electrostatic potential surface analysis shows a slight preference for attack on the outer sulfur.¹⁰³

Steric hindrance is a critical factor that can bias nucleophilic attack toward the outer sulfur, while unhindered persulfides can be attacked on the inner sulfur releasing H₂S.^495,543 The importance of steric hindrance has been documented by comparing the base-promoted decay of trityl- and adamantylpersulfides versus benzylpersulfide. In the former cases, thiol and S_n were the reaction products, while polysulfides (RSSnSR) and H₂S were formed with the more accessible benzylpersulfide.^494,495

Another factor that critically influences the site of attack is the acidity of the leaving group. H₂S has a pK_a of 6.98, while thiols have higher pK_a values (8.29 and 8.94 for cysteine and glutathione, respectively, 25 °C).⁵⁴⁴ It is therefore expected that nucleophiles will attack cysteine persulfides at the inner sulfur releasing the sulfide anion (HS⁻), which is the better leaving group. With protein persulfides, geometric considerations around the persulfide also determine the electrophilicity of the two sulfur atoms. For example, in MST, attack on the outer sulfur is favored by steric and inductive effects, which are governed by active site residues that also render the persulfide highly electrophilic.²⁸⁰

Persulfides can also react with substituted phosphines (R₃P) in a process that is reminiscent of hydroperoxides. The major products formed are phosphine sulfide (R₃P═S) and thiol.⁴⁹⁶ Analysis of the reaction products provided evidence that, while the attack can occur at either sulfur atom, attack at the outer sulfur predominates, particularly when the persulfide is sterically hindered.⁵⁴⁵ Phosphines have been used to detect persulfides in biological samples (see section 7.3.2).

The reaction of persulfides with cyanide to give thiols and thiocyanate (eq 50) deserves mention.

$$\text{RSSH}+{\text{CN}}^{-}\to \text{RSH}+{\text{SCN}}^{-}$$ (50)

This reaction can be used for the detection of persulfides (see section 7.3.2) since thiocyanate can react with ferric ions forming a red complex that absorbs at 460 nm and can be quantified spectrophotometrically (eq 51).³¹

$${\text{SCN}}^{-}+{\text{Fe}}^{3+}\to {\text{FeSCN}}^{2+}(\text{colored})$$ (51)

The reaction of persulfides with cyanide is favored at pH 7.4 relative to pH 10, suggesting that RSSH is the reactive species.¹⁰³ From a mechanistic standpoint, cyanide has been postulated to react with the RS(S)H tautomer of persulfide.⁵⁴⁶ However, computational modeling of the reaction of methylpersulfide (CH₃SSH) with cyanide in a polar medium predicted a linear transition state with an ∼50 kJ mol⁻¹ activation free energy barrier, which is compatible with a nucleophilic displacement mechanism. While the activation free energy barrier for the nucleophilic attack on the inner sulfur was 3 kJ mol⁻¹ lower than the attack on the outer sulfur, the free energy change for the reaction was higher by 96 kJ mol⁻¹, indicating that the reaction on the outer sulfur is thermodynamically favored.¹⁰³ MST and rhodanese catalyze transfer of the outer sulfur from their active site cysteine persulfides to cyanide and to other acceptors (thiols, sulfite). In addition, thiols, which are present at millimolar concentrations inside cells,⁵⁴⁷ are likely to react with LMW persulfides.

Reaction with a thiol at the inner sulfur can give rise to disulfide and H₂S (eq 52).

$$\text{RSSH}+\mathrm{R}\prime {\mathrm{S}}^{-}\rightleftharpoons \text{RSSR}\prime +{\text{HS}}^{-}$$ (52)

Penicillamine-derived persulfide reacts with glutathione to generate H₂S,⁵⁰¹ while glutathione peroxidase-3 persulfide formed a mixed disulfide between the protein and the thiol in the reactions with glutathione and N-acetyl cysteine.⁵⁰⁹ The general reaction described in eq 52 provides a mechanism for H₂S generation from persulfide (see section 8.9).

Reaction at the outer sulfur with a LMW or protein thiol results in trans-persulfidation (eq 53). This reaction is relevant for some proteins involved in iron–sulfur cluster formation and H₂S biosynthesis and oxidation. The role of trans-persulfidation in protein persulfide formation and signaling, and in enzyme-catalyzed depersulfidation is discussed in section 8.

$$\text{RSSH}+\mathrm{R}\prime {\mathrm{S}}^{-}\rightleftharpoons {\text{RS}}^{-}+\mathrm{R}\prime \text{SSh}$$ (53)

7.2.6. Spontaneous Decay of Persulfides

Persulfides are unstable in aqueous solution, which poses challenges for their characterization. For example, real-time MS analysis of penicillamine-derived persulfide showed that it decays with a half-life of 2.7 min at 23 °C;¹³⁵ somewhat higher values have been reported for the decay of CysSSH (35 min at 37 °C).²²⁶ The decay represents a disproportionation reaction involving two molecules of persulfide (eqs 54–56),⁵⁴² which is consistent with the sulfur atoms possessing both electrophilic and nucleophilic character. The importance of RSSH ionization is evidenced by the dependence of the decay rate of persulfides in organic solvents on the strength of the added base.⁴⁹⁴ Acidic conditions also appear to favor decay.^495,499

$$\text{RSSH}+{\text{RSS}}^{-}\rightleftharpoons {\text{RSSS}}^{-}+\text{RSH}$$ (54)

$$\text{RSSSH}\to \text{RSH}+{\mathrm{S}}_2$$ (55)

$$\text{RSSH}+{\text{RSS}}^{-}\rightleftharpoons \text{RSSSR}+{\text{HS}}^{-}$$ (56)

The decay products vary depending on the site of the original attack. Persulfides with bulky substituents react preferentially at the outer sulfur yielding thiol and elemental sulfur (eqs 54 and 55), while those with small substitutents react at the inner sulfur yielding polysulfanes and H₂S (eq 56).^{226,494,495,501} Attack at the inner sulfur is also favored by the release of HS⁻, which is a better leaving group than RS⁻ as discussed previously. Cysteine persulfide predominantly decays via the reaction shown in eq 56,²²⁶ and a similar behavior has been reported for the penicillamine-derived persulfide.⁵⁰¹

In addition to disproportionation, persulfides can undergo thermal or light-induced homolysis of the S–S bond giving the corresponding RS^• and HS^• radicals.⁵⁴⁸ This behavior is expected from the S–S bond energies of HS-SH (276 kJ mol⁻¹) and CH₃S-SCH₃ (309 kJ mol⁻¹).⁵²⁹ Homolysis of persulfides is very slow and is unlikely to contribute to their decay at room temperature and under moderate light.⁵⁴⁸

7.3. Methods for Detecting Persulfidated Cysteines

Spectrophotometric analysis of protein persulfides is of limited utility as they absorb between 335 and 340 nm and have a weak extinction coefficient (∼300 M⁻¹ s⁻¹).^31,549 The IR spectra of thiols and persulfides are similar. However, a band in the 200–500 cm⁻¹ region due to the S–S bond is characteristic of persulfides (Table 3).^491–494 With a few exceptions (Figure 17A) these spectroscopic methods have limited applicability for detecting protein persulfides in complex mixtures.

Figure 17.

Methodological approaches for the characterization of protein persulfides. (A) When pure, protein persulfides can be characterized by UV–visible, IR, and ¹H NMR spectroscopy. (B) Protein persulfides react with 1-fluoro-2,4-dinitrobenzene to form mixed disulfides. DTT releases 2,4-dinitrobenzenethiol, which absorbs at 408 nm under alkaline conditions. (C) Persulfides can be labeled with thiol blocking reagents such as iodoacetamide and analyzed by MS. (D) Protein persulfides can be tagged through different strategies that rely on either their electrophilic or their nucleophilic character.

The major challenge in developing labeling methods is to discriminate between the reactivity of persulfides and thiols, disulfides, sulfenic acids, and polysulfides. Consequently, not too many reliable methods are currently available. The available methods for intracellular persulfide detection can be grouped into two categories: (i) methods for protein persulfide labeling (which rely on the nucleophilic nature of the outer sulfur) and (ii) methods for sulfane sulfur detection (which rely on the electrophilic nature of the persulfide).

7.3.1. Methods for Protein Persulfide Labeling

Due to their greater nucleophilicity, persulfides react faster with commonly used thiol blocking electrophiles than the corresponding thiols⁶⁰ and yield distinct products. Thus, alkylation of thiols yields thioethers while disulfides are formed from persulfides.⁵⁰⁹ Several methods exploit these characteristics of persulfides for their detection.

A spectroscopic method for the indirect detection of protein persulfides relies on the reaction of protein persulfides with 1-fluoro-2,4-dinitrobenzene to form mixed disulfides.⁵⁵⁰ Subsequent treatment with 1,4-dithiothreitol (DTT) releases 2,4-dinitrobenzenethiol with a characteristic absorbance at 408 nm under alkaline (1 M NaOH) conditions. Using an extinction coefficient of 13 800 M⁻¹ cm⁻¹ for 2,4-dinitrobenzenethiol, the protein persulfide concentration can be estimated (Figure 17B).⁵⁵⁰

MS can be used to directly detect the presence of persulfides in proteins.¹⁹ However, the mass increase due to the addition of one sulfur atom (Δm/z = 31.97207) is very similar to that caused by the addition of two oxygen atoms (Δm/z = 31.98984) and can only be distinguished in small peptides but not whole proteins.¹⁹ The relative instability of persulfides further limits their direct detection by MS analysis. To circumvent these problems, persulfidated proteins can be blocked with agents such as N-ethylmaleimide or iodoacetamide (Figure 17C) and then analyzed by MS.^505,509 This treatment stabilizes the persulfide modification, and the detection of a Δm/z of 32 with respect to the nonpersulfidated peptide, confirmed by daughter ions, provides strong evidence for the presence of persulfide.

In principle, treatment of cells with H₂³⁵S followed by Western blot analysis and radioactivity measurement can provide a semiquantitive estimate of protein persulfidation levels. However, the inavailability of H₂³⁵S and the instability of persulfides limit this approach.

The modified biotin switch method first used for proteomic analysis of persulfides¹⁹ was based on the premise that, unlike thiols, persulfides would not react with the electrophilic thiol-blocking reagent, S-methylmethanethiosulfonate (MMTS). The strategy involved initial blocking of thiols with MMTS followed by persulfides labeling with N-[6-(biotinamido)hexyl]–3′-(2′-pyridyldithio)propionamide (biotin-HPDP; Figure 18A). One problem with this method is that MMTS treatment induced intra- and intermolecular protein disulfides.⁵⁵¹ However, the more important flaw with the modified biotin swtich approach for persulfide proteome mapping is that MMTS and its analogue S-4-bromobenzyl methanethiosulfonate react readily with protein persulfides.⁵⁰⁹ Unfortunately, despite these limitations the modified biotin switch method continues to be used and has been shown to illustrate decreased labeling in CSE^19,552–554 and CBS^555,556 knockout cell lines and tissues and increased labeling in response to H₂S treatment. In some cases, cysteines identified as persulfidated targets have been validated by mutagenesis.^552,554,557 A plausible explanation for how some persulfides might become labeled by the modified biotin-switch assay is that persulfides react with MMTS faster than free thiols⁵⁰⁹ and the resulting RS–S–S–Me reacts with residual free thiols regenerating free thiols at sites that carried the persulfide modification originally. The newly formed thiol is subsequently labeled by biotin-HPDP or by fluorescently tagged maleimides (Figure 18B). Given the chemical issues in its design and the resulting misrepresentation of the persulfide proteome, the use of the modified biotin switch method is discouraged.

Figure 18.

Modified biotin switch assay for persulfide labeling. (A) MMTS was proposed to selectively block free thiols leaving persulfides unmodified and ready for reaction with a biotin-derivatized reactive disulfide (biotin-HPDP). (B) Mechanistic explanation for persulfide labeling with the biotin switch assay. MMTS reacts with persulfides more readily than with thiols forming a trisulfide product. This trisulfide is attacked by unreacted thiols leaving a free cysteine that can react with biotin-HPDP.

A second method for persulfide detection involves blocking free thiols and persulfides with Cy5-maleimide and subsequently reducing the R–S–S–maleimideCy5 adduct (Figure 19),⁵⁵² which results in the loss fluorescence. The loss of red fluoresecence is detected following separation of proteins by gel electrophoresis. While relatively simple, the limitation of this method is that it is based on the absence of a signal associated with persulfides rather than on a positive signal, which can be coupled to MS for proteomic analysis. Furthermore, since maleimides are known to react with amines, extensive labeling can give high backgrounds obscuring changes in signal intensity when persulfides are present at low concentrations.

Figure 19.

Persulfide detection by differential fluorescence tagging. Both thiols and persulfides are initially blocked with Cy5-maleimide. DTT treatment then removes the fluorescent tag from the persulfides. Proteins are separated by electrophoresis, and the loss of fluorescence caused by DTT is used as a measure of persulfidation.

A method that allows trapping of persulfidated proteins and subsequent MS analysis⁶⁰ is the basis of the biotin thiol assay (BTA).⁵⁵⁸ Biotin maleimide⁶⁰ (or maleimide-PEG₂-biotin)⁵⁵⁸ is initially used to react with thiols and persulfides (Figure 20). Proteins are then trypsinized, and thiol- and persulfide-tagged biotinylated peptides are immobilized on streptavidin beads. Persulfidated peptides are reduced and peptides are eluted from the beads and derivatized with iodoacetamide for subsequent MS analysis. The use of heavy (deuterated) and light iodoacetamide allows for quantitative analysis.⁵⁵⁸ Using the BTA method >800 proteins have been identified as persulfidation targets.⁵⁵⁸ Modifications of this method referred to as qPerS-SID⁵⁵⁹ and ProPerDP⁵¹⁴ have been reported in which biotinylated iodoacetamide is used instead of biotin maleimide and (tris(2-carboxyethyl)phosphine) instead of DTT.

Figure 20.

Persulfide labeling with biotin-tagged alkylating reagents. (A) In the first step proteins are mixed with maleimide-biotin (or maleimide-PEG₂-biotin) to tag both thiols and persulfides. Proteins are trypsinized and the biotinylated peptides are bound to streptavidin beads. Persulfidated peptides attach to streptavidin beads via disulfide bonds. DTT treatment facilitates elution from the beads and subsequent MS analysis. (B) A possible caveat is that peptides connected by disulfide bonds and containing a thiol or persulfide could be released from the beads with DTT. However, the concentration of disulfide bonds in intracellular proteins is low and this not expected to be a quantitiatvely major drawback of the method.

One difference between the ProPerDP and BTA method is that in the former intact proteins rather than peptides are eluted from the streptavidin beads (Figure 21). Immobilization of intact proteins on streptavidin beads is, however, not recommended since it leads to an underestimation of persulfide targets. For example, a protein containing three cysteine residues (e.g., GAPDH) of which only one is persulfidated will be eluted with a lower yield as binding can occur via any of the cysteine sites. This could explain the relatively low number of protein persulfidation targets identified by the ProPerDP method (Figure 21).⁵¹⁴

Figure 21.

Caveats of whole protein labeling with biotin-tagged alkylating reagents. Proteins containing several cysteine residues, of which only one is persulfidated, are labeled with biotin-maleimide (biotin-NEM) or biotin iodoacetamide (IAB). Since all labels have equal chances of binding to streptavidin beads, the elution with DTT will result in lower than expected yield because the protein will remain bound through the tagged thiols.

Combining the persulfide labeling qPerS-SID approach with the SILAC (stable isotope labeling with amino acid in culture) method allows for quanititative persulfide proteomics analysis⁵⁵⁹ (Figure 22). However, the selectivity of the qPerS-SID and similar methods would be improved by optimizing the initial blocking conditions. As described for the BTA method,⁵⁵⁸ limiting the concentration of the blocking reagent and the duration of labeling, decreases background labeling and identification of false positives (e.g., due to the reactivity of sulfenic acids with IA-biotin⁵⁶⁰). A potential limitation of persulfide tagging methods is that the reduction step also reduces disulfide bonds that were originally present within and across proteins. These disulfide-containing peptides/proteins will, however, only be released from streptavidin beads if they contain a biotin-tagged persulfide but not a biotin-tagged thiol (Figure 20B). In the former case, the identity of the cysteine that was the site of persulfidation will not be revealed by the MS/MS analysis. However, disulfides are not common in intracellular proteins, and the interference by disulfide-containing peptides/proteins in persulfide identification is expected to be low.

Figure 22.

qPerS-SID (quantitative persulfide site identification) approach. Control cells and H₂S-treated cells are grown in heavy and light SILAC (stable isotope labeling with amino acids in cell culture) media, respectively. After cell lysis protein extracts are mixed 1:1 and exposed to iodoacetamide-PEG-biotin to labelthiols and persulfides. Proteins are then trypsinized and labeled peptides bound to streptavidinbeads. Since persulfidated peptides attach to streptavidin beads via disulfide bonds they are eluted with TCEP (tris(2-carboxyethyl)phosphine). Released peptides are subjected to LC-MS/MS identification and quantification (enabled by the SILAC approach).

A related approach that was used to detect tyrosine phosphatase 1B (PTP1B) persulfidation⁵⁶¹ used iodoacetamide to initially block free thiols and persulfidated cysteines, followed by reduction and capture of the newly exposed cysteine thiol with biotinylated iodoacetamide. Although potentially useful to identify potential persulfidation sites on purified proteins, DTT treatment also reduces other oxidative cysteine modifications that might exist on the protein (e.g., nitrosothiol and sulfenic acid) confounding the result.

In another approach, thiols and persulfides are blocked with a maleimide derivative with a peptide arm (MalP).⁵⁶² The persulfidated protein is then detected directly in gel upon loss of the succinimide-peptide moiety (the product of maleimide reaction with a sulfhydryl) upon cleavage of the S–S bond by DTT. The size of the MalP derivative was optimized to cause a detectable shift in protein migration by denaturing polyacrylamide gel electrophoresis.⁵⁶² The best results were obtained with a 16-mer of MalP (MalP16:1.95 kDa; Figure 23). This method is useful for monitoring persulfidation changes in a known target but not for proteome wide analysis.

Figure 23.

Detection of protein persulfidation by differential peptide tagging. An engineered maleimide with a peptide arm and a molecular mass of ∼2 kDa (MalP) reacts with both thiols and persulfides. DTT removes the mixed disulfides formed with persulfides and increases the electrophoretic mobility of the protein.

An alternative tag-switch strategy for identifying protein persulfides is based on the premise that the disulfide bonds resulting from alkylation of protein persulfides with an appropriate thiol blocking reagent show enhanced reactivity to nucleophiles than protein disulfides in which the electrophilicity of the two sulfur atoms is similar (Figure 24).^135,511,524 Therefore, it is possible to introduce a tag-switching reagent (containing both the nucleophile and a reporting molecule) to label only the persulfide products. It should be noted that thiol products are thioethers, which are not expected to react with the nucleophile. Typical thiol-blocking reagents such as maleimides and iodoacetamides are not suited for the tagswitch technology. However, a reagent that gives a mixed aromatic disulfide linkage when reacting with persulfides provides the differential reactivity criteria for the tag-switch technology. In the first step, MSBT or its more water-soluble analogue (benzothiazole-2-sulfonyl)-acetic acid (MSBT-A)⁵⁶³ is used to block thiols and persulfides. In the next step, a biotinylated derivative of methyl cyanoacetate serves as a nucleophile to label the inner sulfur atom of the benzothiazole-blocked persulfide. The selectivity of this method was demonstrated by the reactivity of persulfidated but not glutathionylated, sulfenylated or unmodified bovine serum albumin, which contains intramolecular disulfides in addition to a reactive cysteine.^511,524 The sensitivity of this method has been increased with two new cyanoacetic acid derivatives containing the fluorescent BODIPY moiety (CN-BOT) or the Cy3-dye (CN-Cy3; Figure 24B). These probes allow detection of persulfidated proteins by fluorescence confocal microscopy or in gels.¹³⁵ The reactivity of sulfenic acids with cyanoacetic acid-derivatives is a potential concern and this can be prevented by blocking sulfenic acids with dimedone prior to the reaction with MSBT,^511,524 although no difference in detected persulfide levels was observed between dimedone pretreated and untreated samples.^511,524 The concern that highly reactive protein disulfides would cross-react with cyanoacetic acid-derivatives is overcome by the denaturing conditions of the assay in which the different protein benzothiazole derivatives would be expected to show similar reactivity. The method has been successfully used for mining of the persulfidation proteome in Arabidopsis thaliana, identifying >2000 persulfidated protein targets (5% of the entire Arabidopsis proteome).⁵⁶⁴

Figure 24.

Cyanoacetic acid-based tag-switch method for persulfide labeling. (A) Both persulfides and thiols are initially blocked with MSBT. The product with a persulfidated cysteine is a mixed aromatic disulfide that can be nucleophilically attacked by a cyanoacetic acid–based probe causing a tag-switch. (B) Three different tags are attached to cyanoacetic acid: BODIPY (CN-BOT), Cy3 (CN-Cy3), and biotin (CN-biotin).

7.3.2. Methods for Sulfane Sulfur Detection

The simplest way to detect the total sulfane sulfur pool is by reducing the sample with DTT (Figure 25),^100,101 which releases H₂S. The latter can then be detected by one of several methods as discussed in section 3. However, this method can detect sulfurs from iron sulfur clusters, in addition to inorganic polysulfides and thiosulfates.⁵⁶⁵

Figure 25.

Overview of methodological approaches for total sulfane sulfur detection. Sulfane sulfur compounds release H₂S upon DTT treatment. Sulfane sulfur can be detected by the cold cyanolysis method; samples are incubated in alkaline (pH 8–10) cyanide solutions to release SCN⁻, which is then quantified spectrophotometrically as a complex with Fe³⁺. Sulfane sulfur can be extracted by triarylphosphines in the form of triarylphosphine sulfide, which can be quantified by isotope dilution MS analysis. Finally, different fluorescence probes (e.g., the SSP series) can detect sulfane sulfur in protein persulfides and low molecultar weight hydropolysulfides in cells (described in greater detail in Chart 28).

Cold cyanolysis is widely used to detect sulfane sulfur (Figure 25).^{31,60,566,567} In this method, the cyanide anion attacks the sulfur–sulfur bond^103,546 in persulfides, polysulfides (RSS_nR), and aryl thiosulfonates (pH 8.5–10, 10 °C to room temperature). The resulting thiocyanate is converted to ferric thiocyanate and measured by its characteristic absorbance at 460 nm. When the reaction is performed at higher temperatures (referred to as “hot cyanolysis”), sulfane sufur from thrithionate (⁻O₃SS_nSO₃⁻) and thiosulfate (⁻SSO₃⁻) can be detected as well.³¹ The reactivity of sulfane sufur toward cyanolysis decreases in the following order: persulfide > polysulfide > thiosulfonate > polythionate = thiosulfate > elemental sulfur.

Fluorescent probes have been developed to visualize and quantify sulfane sufur (Figure 25) in cells. The first such probes described were SSP1 and SSP2 (Chart 28A).⁵⁶⁸ Sulfane sufur reacts with a nucleophilic thiol in the nonfluorescent SSP1/SSP2 probe to form a persulfide intermediate, which in turn reacts with an electrophilic ester group leading to spontaneous cyclization and release of the fluorophore (Chart 28B). Negligible basal fluorescence is seen in cells treated with SSP1 or SSP2 indicating that they are either inefficient at detecting protein and/or LMW persulfides or that the concentrations of these compounds are very low.

Chart 28. Fluorescent Probes for Sulfane Sulfur^a.

^a(A) Internal cyclization and fluorophore release following the initial reaction of the probe with sulfane sulfur compounds. (B) Structures of the synthetic probes that exhibit this type of chemical reactivity.

Electrophilic probes like DSPI-3, which initially reacts with inorganic polysulfides and releases the fluorophore upon internal cyclization, have been reported.^569–571 (Chart 29). While DSPI-3 reacts readily with thiols as well, the subsequent intramolecular cyclization leading to generation of a fluorescent signal does not occur (Chart 29A). However, if the thiol adduct were to react further with polysulfides, fluorophore release would occur.⁵⁶⁹ The reaction of protein persulfides or LMW persulfides with DSP probes has not been tested. Optimization of this class of probes for greater selectivity for polysulfide has been reported together with the development of a FRET probe, which detects both H₂S and polysulfides (Chart 29C).⁵⁷¹

Chart 29. Electrophilic Probes that Release a Fluorophore upon Reaction with Inorganic Polysulfides^a.

^a(A) Reactions leading to fluorophore release. (B) Structures of the synthetic probes that exhibit this type of reactivity. (C) FRET-based probe for the simultaneous detection of H₂S and sulfane sulfur-containing species.

A ratiometric near-IR fluorescence probe (Cy-Dise) for cysteine persulfide based on a selenium–sulfur exchange reaction has been reported (Chart 30A).⁵⁷² The method exploits the lower pK_a and greater nucleophilicity of cysteine persulfide compared to cysteine thiol. It is unclear however how this probe is selective for cysteine persulfide versus cysteine thiols with low pK_as, other persulfides, or inorganic polysulfides. The same limitation is also true with a persulfide probe that combines nucleophile-induced xanthene fluorescence quenching with coumarin as a FRET donor (Chart 30B).⁵⁷³

Chart 30.

Ratiometric Near-IR Fluorescence Probe for Cysteine Persulfide Detection (A) and FRET Probe Designed for Persulfide Detection (B)

Isotope dilution MS is a reliable method for quantifying total sulfane sulfur pool⁵⁷⁴ and is based on their known reactivity with triphenylphosphines (Figure 25). This method relies on the use of ¹³C isotope-labeled triarylphosphine sulfide as an internal standard spiked into the biological sample treated with triarylphosphine. From the ratio of the MS signal intensities of the internal standard and triarylphosphine sulfide, an estimate of the total sulfane concentration is obtained (Chart 31).⁵⁷⁴

Chart 31. Isotope Dilution Mass Spectrometry Approach for the Detection of Sulfane Sulfur Using Substituted Phosphines^a.

^aThe reaction of triarylphosphine (P2) with sulfane sulfur compounds. After incubation of P2 with cell or tissue samples to form PS2, PS1 (¹³C-labeled triarylphosphine sulfide) is added as internal standard and the samples analyzed by MS.

8. CELLULAR MECHANISMS OF PERSULFIDE FORMATION AND REMOVAL

An initial estimate based on the modified biotin tag switch assay was that up to 25% of all proteins are persulfidated.¹⁹ The BTA method identified 834 persulfidated proteins in a pancreatic beta cell line,⁵⁵⁸ representing ~5% of the proteome. The cyanoacetate-based tag-switch method confirmed that ~5% of the entire plant proteome is persulfidated.⁵⁶⁴ Significantly lower steady-state protein persulfide levels were reported in HEK293 cells (0.15% of the proteome) and in murine liver (1.2% of the proteome) using the ProPerDP method.⁵¹⁴ Unexpectedly high concentrations of LMW persulfides (150 μM GSSH in brain and 1–4 μM cysteine persulfide in different mouse tissues) were reported in an MS study.¹⁶⁵ The isotope-dilution MS method yielded estimates of sulfane sulfur levels in murine plasma and erythrocytes of 4.7–13.1 and 2.3–3.7 nmol/g protein, respectively, and higher values in other organs: 57.0 (liver), 150.9 (kidney), 46.0 (brain), 61.8 (heart), 56.1 (spleen), and 20.8 (lung) nmol/g tissue.⁵⁷⁴ Extrapolating from these values, this study suggests that the sulfane sulfur concentration in plasma and tissues is in the low micromolar range. In the next section, routes for persulfide formation and removal are described.

8.1. Reaction between H₂S and Disulfides

Early reports indicated that cystine and other LMW and protein disulfides react at alkaline pH with sodium sulfide. The product absorbed at 320–350 nm, was unstable, reacted with cyanide, and was assigned as persulfide (eq 57).^502,575,576

$${\text{HS}}^{-}+\text{RSS}{\mathrm{R}}^{\prime }\rightleftharpoons \text{RSSH}+{{\mathrm{R}}^{\prime }\mathrm{S}}^{-}$$ (57)

Thermochemical calculations and computational modeling of the reaction with cystine suggest that the formation of the RSSH and RS⁻ products is uphill by +56.1 kJ mol⁻¹, but that fast equilibration to RSS⁻ and RSH due to the lower pK_a of the persulfide drives the reaction in the forward direction. The reaction is also driven by further reactions of the unstable persulfide product.^60,64

The reactions of H₂S with typical LMW disulfides are slow.^{60,179,503–505} For example, the rate constant for the reaction of H₂S with cystine at pH 7.4 and 25 °C is 0.6 M⁻¹ s⁻¹.⁶⁰ The reaction of H₂S with GSSG is also slow and reversible and leads to GSSH and a mixture of products.^504,505 Nevertheless, the protein environment can accelerate the reaction. For example, in SQR, the reaction of H₂S with an active site disulfide is accelerated by ~10⁶-fold with respect to free cystine.^306,310,311

Reaction of H₂S with typical disulfides is slower than the analogous reaction of RSH (the thiol–disulfide exchange reactions), probably due to the lack of inductive/field or solvation effects attributed to the adjacent methylene group in thiolates.⁶⁰ The logarithm of the pH-independent rate constant for the reaction of HS⁻ with disulfides decreases linearly with a slope of −0.75 as the pK_a of the thiol that constitutes the disulfide increases, consistent with acidic thiols being better leaving groups. Computational modeling predicts a linear transition state as expected for a concerted S_N2 mechanism.⁶⁰ The influence of the thiol pK_a on kinetics supports the prediction that proteins that contain disulfides formed with acidic thiols are better targets for H₂S. In asymmetrical disulfides (RSSR′) as found in proteins the more acidic thiol can be expected to be the leaving group although steric constraints would also influence which sulfur was attacked.

In the cytosol, the concentration of LMW and protein disulfides is very low.^547,577,578 Hence, the direct reaction between H₂S and disulfides could be more relevant in compartments such as the endoplasmic reticulum and under oxidizing conditions.⁵¹¹

8.2. Reaction between H₂S and Sulfenic Acids

Sulfenic acids (RSOH) are usually formed by reaction of a thiol with a hydroperoxide or with hypochlorous acid.⁵⁷⁹ The sulfur in sulfenic acid is a weak nucleophile and also a soft electrophile.^580–582 Sulfenic acids are typically unstable and decay mainly by reaction with thiols forming disulfides.^581,582 In addition sulfenic acids can react with H₂S (eq 58).

$${\text{HS}}^{-}+\text{RSOH}\to \text{RSSH}+{\text{OH}}^{-}$$ (58)

The formation of persulfide in this reaction was confirmed using the sulfenic acid formed on Cys34 of human albumin.^60,511 This sulfenic acid is located in a cleft with no neighboring thiols. The second-order rate constant for this reaction is 270 M⁻¹ s⁻¹ at pH 7.4 and 25 °C.⁶⁰ The pH-independent rate constant with H₂S is ~4-fold higher than with the corresponding thiol⁶⁰ and suggests that steric constraints influence the effective access of the nucleophile.

Intracellular persulfide levels increase when cells are treated with H₂O₂, suggesting the relevance of sulfenic acid for priming the persulfide modification on proteins. Persulfide levels in H₂O₂-treated cells were decreased by inhibiting CBS and CSE.⁶⁰ Since H₂O₂ can also promote disulfide formation in cells, the effect of diamide, which only supports disufide formation, was compared to the effect of H₂O₂. However, diamide had the opposite effect, leading to lower persulfide levels, which is consistent with the kinetic data that the reaction between disulfides and H₂S is slow and therefore unlikely to be physiologically significant except in special cases.¹³⁵ Under conditions of endoplasmic reticulum stress,⁵⁵⁸ which leads to enhanced ROS production,⁵⁸³ increased persulfidation was observed.

Sulfenic acid can be further oxidized to sulfinic acid (RSO₂H) and sulfonic acid (RSO₃H)⁵⁸⁰ which are typically irreversible modifications.^584–587 As discussed previously in section 7.2.4, reaction of H₂S with protein sulfenic acids to form persulfides could protect against overoxidation and irreversible protein damage. Namely, oxidation of persulfidated residues would result in the formation of RSSO₂H and RSSO₃H, which could be reduced back to thiols in the cell.

8.3. Reaction between H₂S and S-Nitrosothiols as a Source of Persulfides

S-Nitrosothiols represent another post-translational modification of cysteine residues important for the regulation of protein function,^432–437 as discussed in section 6.3. When reacting with thiols, S-nitrosothiols usually undergo trans-nitrosation, a reaction that is largely thermoneutral (eq 59).⁴³⁷

$$\text{RSNO}+{{\mathrm{R}}^{\prime }\mathrm{S}}^{-}\rightleftharpoons {\text{RS}}^{-}+{\mathrm{R}}^{\prime }\text{SNO}$$ (59)

An alternative thiol-assisted decomposition of S-nitrosothiols has been proposed in which disulfide and HNO are formed as products.^588,589

$$\text{RSNO}+\text{RSH}\to \text{RSSR}+\text{HNO}$$ (60)

Consistent with this mechanism, an increase in protein S-glutathionylation is seen in cells treated with GSNO.⁵⁹⁰ However, since this reaction is thermodynamically unfavorable (~+30 kJ/mol), it would be relevant only if it were coupled to product removal.^64,449

The expected product for the reaction of H₂S with RSNO is HSNO.^449,453 Formation of HNO and a protein persulfide (eq 61) is thermodynamically unfavored (+26 kJ/mol),⁶⁴ although the protein microenvironment could facilitate this reaction.

$$\text{RSNO}+{\text{HS}}^{-}\to {\text{RSS}}^{-}+\text{HNO}$$ (61)

The dichotomous reactivity of RSNO with nucleophiles can be explained by the unusual electronic structure of the —SNO group.⁵⁹¹ The resonance representations of RSNO include the standard Lewis structure with a single S—N bond, a zwitterionic structure with an S═N double bond (RS⁺═N—O⁻) and an ionic structure I (RS^−…NO⁺) (Figure 26A). Direct interaction with charged and polar residues in the protein microenvironment could affect the electronic structure of the —SNO group by increasing the electrophilicity of the S atom and therefore its susceptibility to nucleophilic attack (eq 61) or by weakening the S—N bond and increasing the electrophilicity of the N atom thereby promoting transnitrosation (eq 59).⁵⁹² In addition, external electric fields created by the protein environment could influence the electronic structure of RSNOs so as to favor thiolation (or persulfidation in case of the reaction with HS⁻) over trans-nitrosation reaction (Figure 26B).^592,593

Figure 26.

Factors that might favor protein persulfidation over trans-nitrosation in the reaction of an S-nitrosothiol (RSNO) with H₂S. (A) Resonance structures of RSNO. Nucleophilic attack on the sulfur is favored in Structure D. (B) Interactions with the protein environment could stabilize the resonance structure D. Positively charged Arg or Lys residues could stabilize the NO moiety, while negatively charged Glu or Asp residues could stabilize the sulfur. In addition electric fields (EF) created by the protein environment could selectively stabilize a resonsance structure, e.g., D, promoting persulfidation and HNO release.

S-Nitrosation and persulfidation might differentially regulate protein function.⁵⁹⁴ In one study, the persulfide and S-nitrosothiol proteomes reportedly exhibited a 36% overlap.⁵⁵⁸ Further work is needed to delineate the effects of these modifications on proteins function.

8.4. Role of Polysulfides in Protein Persulfidation

Inorganic polysulfides (HSS_nS⁻) and polysulfanes (HSS_nSH) are sulfane sulfur compounds and prone to nucleophilic attack by thiolates. The high abundance of contaminating inorganic polysulfides in H₂S solutions (particularly when NaHS is the source) and their chemical reactivity have led to their contributing to the numerous effects originally ascribed to H₂S. For instance, at nanomolar concentrations, polysulfides activate TRPA1⁵⁹⁵ while H₂S at micromolar concentrations fails to do so.^267,595 Similarly, polysulfide contaminants in an NaHS-derived solution led to disulfide bond formation in the lipid phosphatase and tensin homologue (PTEN),⁵⁹⁶ by attack of a proximal cysteine on the initially polythiolated cysteine (Chart 32A). Inorganic polysulfides also contain negatively charged terminal sulfurs that could react with electrophiles. Polysulfides or H₂S solutions containing traces of Fe³⁺ or Cu²⁺, which stimulate polysulfide formation, were shown to reduce disulfide bonds in human immunoglobulins.⁹⁷ Hence, an additional caution with the use of H₂S solutions containing polysulfide contaminants is that they can react with disulfide bonds forming RSH and RSnS⁻ (Chart 32B).

Chart 32. Reactions of Inorganic Polysulfides with Proteins^a.

^a(A) Inorganic polysulfides can react as electrophiles with a protein thiolate. The resulting polythiolated cysteine can have different numbers of S atoms and can in turn, be attacked by a proximal cysteine forming a family of products, e.g., intramolecular disulfides, trisulfides, etc. (B) Inorganic polysulfides can react as nucleophiles with an intramolecular protein disulfide forming thiol and polythiolated cysteine.

While much emphasis has recently been placed on the potential role of inorganic polysulfides as signaling molecules that are responsible for the effects of H₂S,^597,598 it is unclear and unlikely that they serve this role under physiological conditions. Polysulfides are charged and full protonation is almost impossible under physiological conditions, making diffusion across membranes very slow. Considering the very low steady state levels of H₂S and its tightly regulated oxidation (see sections 3.5 and 5), it is difficult to envision that stochastic oxidation could result in significant amounts of inorganic polysulfides. Polysulfides are unstable and can be readily reduced so it is additionally unclear how they endure reducing intracellular environement. Finally, to serve in signaling, synthesis of polysulfide (which represents a family of catenated sulfur compounds of variable length) should be tightly regulated and their action on targets should be exerted with specificity. To date, none of these criteria are met by polysulfides. An enzymatic reaction in mammals that leads to regulated polysulfide synthesis is not known. Instead, polysulfide synthesis appears to be stochasticity controlled by the concentration of oxygen and metals on the one hand and protein thiols/disulfides on the other.

The reported production of polysulfides (H₂S₂ and H₂S₃) from 3-mercaptopyruvate by MST⁵⁹⁹ was problematic for the following reasons. Polysulfide formation from 3-mercaptopyruvate was observed when the reaction was run in the absence of a sulfur acceptor. Under these conditions, the K_M for 3-mercaptopyruvate was reportedly 4.5 mM, although substrate inhibition led to complete loss of enzyme activity above 2 mM concentration.²⁷¹ Furthermore, the K_M value obtained in the H₂S₃ synthesis assay is at least 10-fold higher than the K_M for 3-mercaptopyruvate (20—350 μM) in the presence of acceptors, which leads to H₂S synthesis. Therefore, conditions supporting polysulfide synthesis by MST are unlikely to exist in the cell since MST exhibits a low K_M (2.5 μM) for its physiological acceptor, thioredoxin.²⁷¹

The chemical characteristics of inorganic polysulfides and polysulfanes make them unlikely to be signaling molecules as well.^600,601 The equilibrium between elemental sulfur and aqueous polysulfide at 25 °C was studied either by adding acid to polysulfide solutions until the sulfur precipitated, or by dissolving elemental sulfur in aqueous polysulfide solution until an equilibrium was established. The ratio between the S_n and HS⁻ species was strongly dependent on the alkalinity of the solution.

$${\mathrm{S}}_{\mathrm{n}}+{\text{HS}}^{-}+{\text{OH}}^{-}\rightleftharpoons {{\mathrm{S}}_{\mathrm{n}+1}}^{2-}+{\mathrm{H}}_2\mathrm{O}$$ (62)

Since the polysulfide dianions of different chain-lengths are in an equilibrium (eq 63) that is rapidly established, it is not possible to reliably separate polysulfide species by ion chromatography.⁶⁰³

$$\begin{gathered} {\mathrm{S}}_m{\mathrm{S}}^{2-}+{\text{HS}}^{-}+{\text{OH}}^{-}\rightleftharpoons {\mathrm{S}}_x{\mathrm{S}}^{2-}+{\mathrm{S}}_y{\mathrm{S}}^{2-}+{\mathrm{H}}_2\mathrm{O} \\ m=x+y \end{gathered}$$ (63)

The pK_as were studied using solutions of pure salts of S₂²⁻ to S₅²⁻ and a special streaming apparatus, which mixed polysulfides with HCl and allowed determination of the pH within 10⁻² s.⁶⁰⁴ The short mixing time averted decomposition of the protonated polysulfide into sulfur and monosulfide. Only the pentasulfide did not equilibrate as multiple species upon acidification, which allowed precise determination of its pK_a. The shorter polysulfides disproportionated within the mixing time of the experiment.⁶⁰⁴ S₂²⁻ and S₃²⁻ could not be detected in an aqueous solution of potassium trisulfide which equilibrated to 18% HS⁻, 62% S₄²⁻, and 20% S₅²⁻.⁶⁰⁴ In fact, other groups have confirmed that HS₂⁻ and S₃²⁻ do not exist at detectable concentrations in neutral solutions.^49,605–609 Tetra-sulfide dianion is the predominant species until a pH of 10–11. However, even tetrasulfides disproportionate to give the pentasulfide (eq 64),^{605,606,608,609} although this reaction is slow.

$$4{{\mathrm{S}}_4}^{2-}+{\mathrm{H}}_2\mathrm{O}\to {\text{OH}}^{-}+{\text{SH}}^{-}+3{{\mathrm{S}}_5}^{2-}$$ (64)

Polysulfanes can be prepared by rapid acidification of crude sodium sulfane (Na₂S_n) to produce raw sulfane (H₂S_n). H₂S₃ (as a side product) and H₂S₂ can be collected by fractional distillation of raw sulfane at room temperature and at −80 °C, respectively.^69,70 Sulfanes are liquids that are miscible with carbon disulfide, benzene, tetrachloromethane, and dry diethyl ether. H₂S₂ and H₂S₃ are colorless/pale yellow, but the higher sulfanes are more yellow.^69,70 Using ¹H NMR, the existence of sulfanes with up to 35 sulfur atoms was demonstrated.⁶¹⁰ Exposure of pure sulfanes to air/humidity caused immediate decomposition. In the cases of H₂S₂ and H₂S₃ the decomposition is explosive.⁶⁹

$$8{\mathrm{H}}_2{\mathrm{S}}_2\left(1\right)\to 8{\mathrm{H}}_2\mathrm{S}\left(\mathrm{g}\right)+{\mathrm{S}}_8\left(\mathrm{s}\right)$$ (65)

The pK_a values of H₂S₂ are pK₁ 5.0 and pK₂ 9.7. The pK_a values of H₂S₃ are pK₁ 4.2 and pK₂ 7.5. In fact, ab initio MO calculations confirm that the acidity of sulfanes increases with the number of sulfur atoms in the molecule.⁶¹¹

Instability of aqueous polysulfide solutions (particularly S₂²⁻ to S₄²⁻) is also due to their rapid autoxidation in air at temperatures between 23 and 40 °C forming thiosulfate and elemental sulfur (eq 66).⁶¹² No other sulfur containing species are detected in these reactions.

$${{\mathrm{S}}_{2+\mathrm{x}}}^{2-}+3/2{\mathrm{O}}_2\to {\mathrm{S}}_2{{\mathrm{O}}_3}^{2-}+\mathrm{x}/8{\mathrm{S}}_8$$ (66)

Polysulfides can also react with sulfite in neutral solution to give thiosulfate and HS⁻ (eq 67).⁶¹³

$${{\mathrm{S}}_4}^{2-}+{{\text{HSO}}_3}^{-}\to 3{\mathrm{S}}_2{{\mathrm{O}}_3}^{2-}+{\text{HS}}^{-}+2{\mathrm{H}}^{+}$$ (67)

Studies in which inorganic polysulfides are trapped as organic polysulfanes (e.g., with monobromobimane) in order to establish their intracellular formation need to be viewed with caution. Organic polysulfanes (RSSnSR) are also unstable; even pure substances tend to decompose by equilibration with other chain lengths and by formation of elemental sulfur. These reactions are accelerated by light and heat, and by the presence of nucleophiles.⁶¹⁴ Due to the intrinsic instability of inorganic and organic polysulfanes and polysulfides neither standards nor products used in this methodological approach are stable.

Based on all these chemical characteristics, it is very difficult to envision a biological setting that is conducive to the regulated production of polysullfides or their utilization in signaling in mammalian systems. However, polysulfides with the caveats of their instability noted above might have some pharmacological potential.

8.5. Persulfide Formation via Radical Reactions

Strong one-electron oxidants can react with H₂S and RSH forming HS^• and RS^•, respectively. Rapid free radical recombination between HS^• and RS^• would lead to persulfide formation (eq 68), although this is highly likely to be an insignificant source of persulfides in cells due to the low concentration of the free radicals.

$${\text{HS}}^{•}+{\text{RS}}^{•}\to \text{RSSH}$$ (68)

An alternative and more likely, radical pathway for RSSH formation is via the reaction of HS^• with RS⁻ or, conversely, RS^• with HS⁻, to form the radical anion, RSSH^•−. The latter can react with oxygen forming RSSH and O₂^•− (eqs 69–71).

$${\text{HS}}^{•}+{\text{RS}}^{-}\to {\text{RSSH}}^{•-}$$ (69)

$${\text{RS}}^{•}+{\text{HS}}^{-}\to {\text{RSSH}}^{•-}$$ (70)

$${\text{RSSH}}^{•-}+{\mathrm{O}}_2\to \text{RSSH}+{{\mathrm{O}}_2}^{•-}$$ (71)

The potential importance of free radical reactions in persulfide synthesis was demonstrated by increased protein persufidation in cell lysates treated with metal ions (Fe³⁺ or Cu²⁺) and H₂S.⁵¹¹ Similarly, persulfidation of BSA was strongly induced by treatment with H₂S and a water-soluble heme iron.⁵¹¹ The reactivity of H₂S and RSH with metal ions under aerobic conditions should serve as a caution for handling cell lysates, which could contain higher free metal ion concentrations and in the presence of H₂S could lead to an artifactual increases in persulfidation.

H₂S can react with the oxidized form of several metalloproteins like cytochrome c^97,505 (see section 5.1) leading to their reduction and formation of HS^•. It is possible that, under conditions of increased H₂S production or decreased oxidation (e.g., during hypoxia),⁶¹⁵ ferric cytochrome c is reduced while H₂S is oxidized to HS^• leading to increased protein persulfidation in mitochondria. The role of metalloprotein-assisted HS^• generation in protein persulfidation remains to be investigated.

The chemistry of polysulfides is closely related to that of the radical monoanions S_x^•− (eqs 72 and 73) in that they are in equilibrium with each other.

$${{\mathrm{S}}_4}^{2-}\rightleftharpoons 2{{\mathrm{S}}_2}^{•-}$$ (72)

$${{\mathrm{S}}_6}^{2-}\rightleftharpoons 2{{\mathrm{S}}_3}^{•-}$$ (73)

In solution these radicals are formed by homolytic dissociation of the polysulfide anions, a process that is enhanced in solvents of lower polarity than water and/or by higher temperatures.^600,616 In aqueous solutions at room temperature, the reactions given in eqs 72 and 73 are shifted to the left. However, if the solutions are very dilute, then radical anions could be more abundant. The radical anions would be reactive toward thiols leading to protein polythiolation and O₂^•− formation (eqs 74 and 75).

$${{\mathrm{S}}_{\mathrm{x}}}^{•-}+\text{RSH}\to {\text{RSS}}_{\mathrm{x}}{\mathrm{H}}^{•-}$$ (74)

$${\text{RSS}}_{\mathrm{x}}{\mathrm{H}}^{•-}+{\mathrm{O}}_2\to {{\mathrm{O}}_2}^{•-}+{{\text{RSS}}_{\mathrm{x}}}^{-}+{\mathrm{H}}^{+}$$ (75)

8.6. Reaction between Thiols and Activated Organic Disulfides and Polysulfides

Organosulfur compounds in garlic can react with thiols and serve as a source of persulfides and H₂S. Allicin (diallyl thiosulfinate) is synthesized from alliin after release of alliinase when garlic is crushed. Allicin is rapidly metabolized in aqueous solution to diallyl sulfide, diallyl disulfide, diallyl trisulfide, and ajoene (Chart 33A).^{105,110,111,617}

Chart 33. Active Principles of Garlic and the Mechanisms for LMW Persulfide Generation from Them^a.

^a(A) Allicin (diallyl thiosulfinate) is rapidly metabolized in aqueous solutions into diallyl sulfide (DAS), diallyl disulfide (DADS), diallyl trisulfide (DATS), and ajoene. (B) Glutathione-promoted decomposition of DADS and generation of allylpersulfide, glutathione persulfide (GSSH), and H₂S.

GSH reacts with the active principles in garlic leading to H₂S release.⁶¹⁸ These reactions are facilitated by allyl substituents and by increasing numbers of sulfur atoms. A similar structure–activity correlation has been reported for the cancer-preventative effects of garlic-derived organic polysulfides.⁶¹⁹ In addition to H₂S, LMW persulfides are also formed in the reaction of GSH with organic disulfides and polysulfides resulting from nucleophilic substitution at the α-carbon, yielding S-allyl-glutathione and allyl persulfide (Chart 33B).⁶¹⁸ The latter reacts with GSH releasing H₂S and allyl-glutathione disulfide, which in turn is an additional target for nucleophilic substitution leading to GSSH formation. Diallyltrisulfide and higher order diallylpolysulfides react in a similar manner and have the additional possibility of undergoing a direct nucleophilic attack by GSH at the sulfane sulfur.⁶¹⁸ In addition to GSH, other thiolates in cells can engage in similar chemical reactions with garlic-derived allyl sulfides.

8.7. Elimination Reactions of Disulfides

Disulfides degrade under alkaline conditions giving rise to a variety of products. Base-promoted elimination of disulfides leads to the formation of the corresponding persulfide and dehydroalanine derivative (Chart 34).^576,620,621

Chart 34.

Base-Promoted Persulfide Generation from Cystine

Elimination reactions of disufide substrates (cysteine and homocystine) are catalyzed by CBS and CSE forming the corresponding persulfides^{165,226,622–624} At physiologically relevant substrate concentrations, the contributions of CBS and CSE to Cys-SSH generation is low and homocysteine persulfide synthesis by CSE is negligible (see sections 4.1.3 and 4.3.1).²²⁶ However, under pathological conditions that lead to cystine accumulation, Cys-SSH synthesis by CSE might become a contributing factor to persulfide synthesis.

8.8. Sulfur Transfer Reactions

Sulfurtransferases react with their substrates forming a persulfide intermediate from which the sulfane sulfur is transferred to an acceptor. In addition to MST, rhodanese and SQR that are directly involved in H₂S synthesis or clearance (sections 4 and 5), the cysteine desulfurases catalyze sulfur transfer reactions needed in biosynthetic pathways e.g. synthesis of iron sulfur clusters^625–636 and thionucleosides.^{629,633,634,636} While some sulfurtransferases (e.g cysteine desulfurase) utilize the PLP cofactor, others do not (e.g., MST). These enzyme promote the formation of a Cys-SSH in the active site and can, in principle, transfer the sulfane sulfur to a thiol on a target protein or to a LMW thiol. The role of these enzymes in catalyzing protein persulfidation remains to be determined.

8.9. Reaction between Thiols and Persulfides (Transpersulfidation)

As discussed in section 7.2, nucleophilic attack by thiolates on persulfides occurs predominantly on the inner sulfur with formation of disulfide and release of H₂S (eq 52). This chemistry has been documented with LMW persulfides and with protein persulfide models.^494,501,509 However, proteins such as the sulfurtransferases modulate the reactivity of persulfides such that the outer sulfur is transferred and H₂S is not released (see section 5.2). Very high levels of LMW persulfides in cells, tissue and circulation (50–100 μM) have been reported¹⁶⁵ leading to the suggestion that Cys-SSH and/or GSSH are major biologically relevant transpersulfidating reagents.⁵²⁷ However, high concentrations of reactive persulfides and especially in an oxidizing compartment like blood are unlikely. Cys-SSH and GSSH have been proposed to be major biologically relevant transpersulfidating reagents that can even be transported across the membrane.^165,637 However, the low tissue levels of sulfane sulfur compounds^32,574 and the fact that their reaction with thiolates, which are abundant in cells, favor H₂S release, argue against this possibility.

The transfer of the terminal sulfur from a persulfide to a thiolate, constitutes a transpersulfidation reaction, and has been documented in several proteins. Sulfur-containing cofactors and modified thionucleosides, as well as iron–sulfur clusters obtain their sulfur atom via transpersulfidation reactions.^636,638 It is likely that steric factors, the acidity of the parent thiol and the protein microenvironment determine the predominance of thiolate attack on the outer versus the inner sulfur of the persulfide.

The reaction mechanism of transpersulfidation by the LMW persulfides (eq 53) has been proposed to involve the tautomeric thiosulfoxide species (Chart 35) as the sulfur donor.^33,34,546 The S═S bond in thiosulfoxides can be considered to be either a double⁶³⁹ or “semipolar”⁶⁴⁰ bond depending on the electronegativity of substituents. Nevertheless, computational studies reveal that, although the thiosulfoxide is only 5 kJ/mol less stable than the corresponding persulfide, the energy barrier for tautomerization is very high, i.e., >100 kJ/mol.⁶⁴¹ Hence, LMW thiosulfoxides appear to be unlikely donors in uncatalyzed cellular transpersulfidation reactions.

Chart 35.

Tautomerization of a Persulfide to a Thiosulfoxide as a Postulated Mechanism for Transpersulfidation

Transpersulfidation can, however, occur in the active site of certain enzymes. A computational analysis of the sulfurtransfer reaction²⁸⁰ based on cystal structure of human MST²⁷¹ suggests that the persulfide anion is the sulfur donor to the thiolate acceptor, in a reaction that is facilitated by the increase in electrophilicity of the outer sulfur through multiple hydrogen bonding interactions. Therefore, the active site geometry and electronics favor transfer of the terminal sulfur, i.e. transpersulfidation to either LMW or a protein thiol (Figure 7).

MST transfers the sulfane sulfur to sulfite forming thiosulfate, albeit very inefficiently.⁶⁴² Treatement of red blood cells with the MST substrate, 3-mercaptopyruvate, reportedly inhibited many glycolytic enzymes including hexokinase, phosphofructokinase, pyruvate kinase, glucose-6-phosphate dehydrogenase and 6 phosphogluconate dehydrogenase.⁶⁴³ Transpersulfidation of the sulfane sulfur from MST to those enzymes was suggested but not established as the mechanism of inhibition.⁶⁴³ An increase of persulfidation was reported in cells treated with D-cysteine,¹³⁵ a source of 3-mercaptopyruvate via D-amino acid oxidase.²⁷⁷ The cryo-EM structure of complex I isolated from Yarrowia lipolytica indicates the presence of an additional subunit, which was identified to be an accessory sulfurtransferase subunit. This rhodanese/MST-like subunit was capable of using 3-mercaptopyruvate as a substrate and, in the presence of thiols, released H₂S. The role of this sulfurtransferase subunit on the structure and function of complex I remains to be elucidated.⁶⁴⁴

There are limited examples of rhodanese-catalyzed sulfur transfer to protein acceptors. Although a role for rhodanese in iron sulfur biogenesis had been previously proposed,^645–649 it has since been shown to not be involved in this process. Rhodanese catalyzed sulfur transfer from thiosulfate to malate dehydrogenase as monitored by ³⁵S radiolabel transfer and led to an almost 2-fold increase in activity. Hence, persulfidation could regulate energy metabolism via the citric acid cycle.⁶⁵⁰ In the presence of thiosulfate, bovine rhodanese restored the activity of partially inactivated NADH dehydrogenase, a subunit of complex I.⁶⁵¹

Rhodanese has been identified as a candidate obesity-resistance gene with increased expression in adipocytes being correlated with leaness.³²⁷ Overexpression of rhodanese in adipocytes protected mice from diet-induced obesity and insulin-resistant diabetes and rhodanese-deficient mice showed aggravated development of diabetes.³²⁷ An earlier study had correlated low rhodanese expression with increased whole cell ROS and mitochondrial O₂^•− levels and higher mortality in hemodialysis patients.⁶⁵² The link between these observations and changes in intracellular persulfide levels and whether rhodanese in fact catalyzes protein persulfidation, remains to be elucidated.

The rhodanese homology domain has been identified in ∼500 proteins in the three major evolutionary phyla. In the human genome, there are 47 examples of rhodanese-domain containing proteins.³²⁸ Cdc25 phosphatase, an activator of cell division kinases during the cell cycle, is an example of rhodanese domain-containing protein. The crystal structure of the catalytic domain of human Cdc25A reveals a small α/β domain with a rhodanese domain fold.⁶⁵³ It is not known, however, whether the signaling role of Cdc25 involves transpersulfidation. Other proteins that have a rhodanese domain and form an active site Cys-SSH are adenylyltransferase and the MOCS3 sulfurtransferase.^654,655 The roles of these proteins in catalyzing transpersulfidation chemistry are, however, likely to be restricted to the specific pathways in which they are involved e.g. molybdopterin biosynthesis.

8.10. Depersulfidation

Signaling via protein persulfidation requires the existence of cellular mechanisms for removal of the persulfide modification and inactivation of the signal. Thioredoxin (Trx) is a 12 kDa disulfide oxidoreductase, which serves as a redox partner for a wide variety of client proteins.^656–659 In humans, Trx1 is present in the cytosol and the nucleus while Trx2 is present in mitochondria. Trx contain two cysteines (Figure 27A) in an active site CXXC motif.⁶⁵⁶ The pK_a values of the nucleophilic cysteine is between 6 and 7 and for the resolving cysteine is between 8 and 9.^660–662 Reduction of a client protein disulfide starts with the attack of the nucleophilic cysteine to form an intermolecular mixed disulfide, followed by subsequent attack of the resolving cysteine on the mixed disulfide to form the fully oxidized Trx. The two-electron reduction potential of the disulfide/dithiol couple in thioredoxin is −284 mV at pH 7.0 and 25 °C.^656,663 The recognition between Trx and its client proteins is postulated to be entropically driven.⁶⁶⁴ Oxidized Trx is reduced by the FAD-containing selenoprotein, thioredoxin reductase (TrxR1). Three isoforms of this protein exist in mammals: cytosolic TrxR1, mitochondrial TrxR2 and TrxR3 present only in testis. TrxR uses NADPH as a source of electrons (Figure 27B). Excellent reviews on the Trx/TrxR system have been published.^656–659

Figure 27.

Depersulfidation by thioredoxin (Trx). (A) The structure of human thioredoxin (PDB: 5DQY). (B) Thioredoxin reduces protein persulfides and releases H₂S. Oxidized Trx is reduced by thioredoxin reductase (TrxR) at the expense of NADPH. (C) Two possible mechanisms for protein depersulfidation. Top: The nucleophilic thiol attacks the inner sulfur of the protein persulfide forming a mixed protein-Trx disulfide and releasing H₂S. In the next step the resolving cysteine reduces the mixed disulfide forming fully oxidized Trx. Bottom: Trx undergoes persulfidation, forming Trx persulfide, which is reduced by the resolving cysteine with concomitant release of H₂S.

As discussed in Section 4.5, the Trx/TrxR is involved in the sulfur transfer reaction catalyzed by MST,^270,271,273 and formation of thioredoxin persulfide was demonstrated with the Trichomonas vaginalis MST.²⁷⁴ Trx is ∼200-fold more efficient at reducing the Cys-SSH in PTP1B than DTT.⁵⁶¹ Addition of Trx to cell lysate resulted in H₂S generation, while the treatment of cells with auranofin, a TrxR inhibitor, increased total intracellular persulfidation levels.¹³⁵ Trx reduced penicillamine-derived HSA persulfide and Cys-SSH.¹³⁵ The first order rate constant for the reaction of Trx with Cys-SSH was estimated to be 4.5 × 10³ M⁻¹ s⁻¹ at 23 °C and pH 7.4, which is almost 10-fold higher than with Cys-SS-Cys. A similar rate constant was observed with HSA-SSH (4.1 × 10³ M⁻¹ s⁻¹).¹³⁵ The Trx/TrxR/NADPH system exhibited Michaelis–Menten-like kinetic behavior. These results are consistent with an important role for the Trx/TrxR system in protein depersulfidation.^135,514 The involvement of a related protein, TRP14 (thioredoxin-related protein of 14 kDa),⁶⁶⁵ was demonstrated by its silencing, which resulted in increased persulfide levels.⁵¹⁴ TRP14 might be important as a depersulfidase particularly under conditions of oxidative stress when Trx is tied up with the peroxiredoxin system.⁵¹⁴

Depersulfidation by Trx can occur via one of two mechanisms: (i) transfer of the outer sulfur from the persulfide to the nucleophilic cysteine of Trx leading to the transient formation of Trx-SSH which is subsequently resolved forming H₂S and oxidized Trx and (ii) a nucleophilic attack on the inner sulfur of the persulfide with elimination of H₂S and formation of a mixed Trx-client disulfide complex, which is resolved (Figure 27C).¹³⁵ While the formation of mixed disulfides is part of the disulfide reductase activity of Trx, the mechanism of the depersulfidase activity remains to be established.

Increased Trx levels are associated with diseases such as rheumatoid arthritis, hepatitis C and HIV-1 infections.^666–668 HIV-1 patients with high viral load have increased levels of circulating Trx.^666,667 An inverse correlation was seen between total plasma sulfane sulfur levels and viral load in HIV-1 patients. Indirectly, this result is consistent with a role for the Trx system in depersulfidation in vivo.¹³⁵

The glutaredoxin system (GSH/glutathione reductase (GR)/glutaredoxin (Grx)) could also be involved in catalyzing depersulfidation in vivo (Chart 36).⁵¹⁴ GR activity has been reported in cytoplasm and in organelles (ER, lysosome, mitochondria and nucleus). GR regulates cellular redox status by maintaining low levels of GSSG and is important in protecting cells from oxidative stress.^669–671

Chart 36. Glutaredoxin-Catalyzed Protein Depersulfidation^a.

^aGlutaredoxin (Grx) reduces a protein persulfide, oxidized Grx is reduced by glutathione (GSH), and GSSG is reduced by glutathione reductase (GR) at the expense of NADPH.

The Grx system efficiently reduced polysulfides and BSA-SSH in vitro. In murine hepatocytes with the double knockout (TrxR1/GR null), increased persulfidation was observed.⁵¹⁴ GR null cells, however, showed no difference in persulfidation levels compared to controls.⁵¹⁴ Since the in vitro assay requires addition of GSH, the observed polysulfide and protein persulfide reduction could have resulted from their direct reduction by GSH.

Beside Grx, the Trx fold is also found in several other classes of enzymes, such as Dsb (disulfide bond formation protein) proteins, glutathione S-transferase, and protein disulfide isomerase (PDI) families.⁶⁵⁶ Further systematic studies should unravel the potential role, if any, of these enzymes in protein depersulfidation.

9. PERSULFIDATION IN ACTION

As a growing list of persulfidated proteins is being identified, the stage is being set for making the connection between these targets and the molecular mechanisms of H₂S action. In this section, we provide an overview of protein targets and downstream signaling pathways that are affected by H₂S-induced persulfidation with the caution that, in many systems, the persulfidation target has not been rigorously established and correlated with functional effects.

9.1. Persulfidation of K_ATP Channels

Although CSE knockout mice exhibit hypertension,¹² which is consistent with a role for H₂S as an endogenous regulator of blood pressure,^7,672 subsequent studies have revealed that this effect is mediated via cross-talk with NO^•,^{266,267,451,673} an established vasodilator. Glibenclamide, a selective potassium ATP channel (K_ATP) blocker, partially inhibits the vasodilatory effects of H₂S.⁶⁷² Persulfidation of Cys43 on the Kir6.1 subunit of the K_ATP channel in smooth muscle cells prevents its association with ATP and promotes binding to phosphathidylinositol-4,5-bisphosphate,⁵⁵³ which leads to channel opening, to K⁺ influx, and, subsequently, to smooth muscle cell relaxation.

9.2. Persulfidation of Keap-1, p66Shc, and RAGE

A major mechanism for upregulating antioxidant enzymes involves activation of the antioxidant response element (ARE) by the oxidative-stress sensor protein Kelch-like ECH-associated protein 1 (Keap1) and the transcription factor nuclear factor (erythroid-derived 2)-like 2 (Nrf-2).^674–676 Normally, Keap1 binds to the Neh2 domain of Nrf-2 and sequesters it in the cytoplasm, where it is targeted for proteosomal degradation. Electrophilic agents such as sulforaphane can modify Keap1, promoting Nrf-2 nuclear accumulation and ARE activation.^674,676 A widely accepted model for the nuclear accumulation of Nrf-2 invokes modification of critical cysteines on Keap-1 resulting in a conformational change, which induces dissociation of the Keap1–Nrf-2 complex leading to nuclear translocation of Nrf-2 (Figure 28).^674–676

Figure 28.

H₂S may regulate cellular antioxidant defenses and prevent senescence by persulfidation of Keap1. In the cytosol, Keap1 represses Nrf-2 signaling by binding to it. Bound Nrf-2 is subjected to polyubiquitination and proteasomal degradation. Persulfidation of cysteine residues in Keap1 induces a conformational change, which results in Nrf-2 release. Nrf-2 translocates to the nucleus where it upregulates the expression of various antioxidant defense genes.

The cardioprotective effects of H₂S, particularly in ischemia-reperfusion injury, depend on the nuclear translocation of Nrf-2 and activation of antioxidant defense enzymes.^677,678 One group has reported that Keap-1 is persulfidated at Cys151 (identified using the modified biotin-switch assay) when cells are exposed to H₂S,⁵⁵⁴ (Figure 28). In contrast, a second study has reported the formation of a disulfide bond between Cys226 and Cys613 in Keap1 in H₂S-treated cells.⁶⁷⁹ Surprisingly, the same study reported that Cys226 and Cys613 were also persulfidated, which is not expected if persulfidation involves attack of the initially formed disulfide by H₂S.⁶⁷⁹ Alternatively, persulfidation could occur by reaction of H₂S with the sulfenic acid derivative of each cysteine. Activation of the Keap1-Nrf-2 signaling cascade resulted in the upregulation of enzymes involved in H₂S metabolism, CBS, CSE, and SQR.⁶⁷⁹

An alternative mechanism proposed for regulating intracellular ROS production involves persulfidation of the p66Shc protein.⁶⁸⁰ The p66Shc protein is a member of the ShcA family with which it shares three functionally identical domains: the C-terminal Src homology 2 domain (SH2), the central collagen homology domain (CH1), and the N-terminal phosphotyrosine-binding domain (PTB).⁶⁸¹ When cells are exposed to oxidative stress caused by exposure to UV light or to H₂O₂, p66Shc is activated by phosphorylation at Ser36. Activated p66Shc is then dephosphorylated and translocated to the mitochondrion, where it binds to cytochrome c and assists in the electron transport process.⁶⁸¹ p66Shc^−/− mice show a 30% increase in lifespan.⁶⁸¹ Persulfidation of p66Shc at Cys59 inhibits its interaction with PKC_βII and attenuates H₂O₂-induced p66Shc phosphorylation, a critical step in p66Shc-mediated mitochondrial ROS generation.⁶⁸⁰ H₂S is known to protect against oxidative stress, which cannot readily be explained by a direct antioxidant role (see section 2). Inhibition of mitochondrial ROS production via persulfidation of p66Shc⁶⁸⁰ and upregulation of antioxidant defense enzymes via Keap1-Nrf-2 signaling^554,679 provide a mechanistic basis for the protective effects of H₂S.

Persulfidation of RAGE (receptor for AGE: advanced glycation end products) at Cys259 and Cys301 by H₂S treatment or CBS overexpression attenuates cell death and senescence caused by both AGE and β-amyloid⁶⁸² AGE are glycated proteins and lipids observed in diabetic patients.⁶⁸³ Dimeric RAGE is processed in the ER and delivered to the cell membrane. Persulfidated RAGE monomers are less stable, which disrupts their translocation from the ER to the plasma membrane and leads to increased protection against cell death and senescence.⁶⁸²

9.3. Persulfidation and ER Stress

ER stress induces a major transcriptional, translational, and metabolic reprogramming in cells and is associated with the development of many diseases, ranging from metabolic dysfunction to neurodegeneration.⁶⁸⁴ ATF4, a master transcriptional regulator, is induced during the ER stress response and upregulates CSE²⁶³ and the cysteine transporter, Slc7a11.⁵⁵⁸ Persuflidation of a number of protein targets is increased during ER stress and is correlated with reprogramming of energy metabolism toward increased glycolytic flux in pancreatic beta cells (Figure 29).⁵⁵⁸

Figure 29.

Possible role of H₂S in endoplasmic reticulum (ER) stress. Under ER stress, the activity of the transcription factor ATF4 is increased, resulting in the upregulation of CSE and the cystine transporter Slc7a11. The subsequent increased production of H₂S leads to the persulfidation of protein tyrosine phosphatase 1B (PTP1B) and consequently to an increase in pERK phosphorylation. pERK activation results in global inhibition of protein translation by activation of eukaryotic translation initiation factor 2α (eIF2α). eIF2α induces ATF4 nuclear translocation. The increased production of H₂S during ER stress also results in persulfidation of glycolytic and tricarboxylic acid (TCA) cycle enzymes.

The ER-stress response also leads to persulfidation of the protein tyrosine phosphatase (PTP) family of enzymes.⁵⁶¹ PTPs are cysteine hydrolases and are sensitive to oxidation.⁶⁸⁵ PTP-1B is a members of this class of enzymes that is located on the cytoplasmic face of the ER and plays an important role in ER stress signaling.⁶⁸⁶ Persulfidation of Cys215 in PTP-1B leads to the loss of enzymatic activity and, consequently, to increased phosphorylation and activation of PERK,⁵⁶¹ which results in global inhibition of protein translation (Figure 29). ER stress conditions induce ROS production, which could promote protein sulfenylation and potentiate subsequent persulfidation.

9.4. H₂S Effects on GAPDH

GAPDH is an important glycolytic enzyme and exhibits high reactivity toward H₂O₂, which oxidizes the nucleophilic Cys152 residue in the active site, inactivating the enzyme. The redox sensitivity of GAPDH is important for metablic adaptation to increased intracellular H₂O₂⁶⁸⁷ levels. S-Nitrosation of the catalytic Cys152 abolishes GAPDH activity and promotes its binding to the E3-ubiquitin-ligase, Siah1.⁶⁸⁸ The latter possesses a nuclear localization tag and leads to nuclear accumulation of GAPDH. Stabilization of Siah1 by GAPDH promotes degradation of nuclear proteins and leads to apoptosis.⁶⁸⁸ Persulfidation of GAPDH reportedly also promotes Siah1 binding although the modification site was not rigorously determined.⁵⁵⁶

Besides the active site Cys152 residue, human GAPDH has two other cysteines, Cys156 and Cys247. Due to its high abundance, GAPDH is a commonly identified target in proteomic searches including persulfide proteomic data sets.^19,558 While some studies have reported activation of GAPDH in response to persulfidation,^19,558 the connection between persulfidation of Cys152 and GAPDH activity has not been rigorously established. Cys152 functions as a nucleophile in the reaction cycle, and given the inhibitory effect of S-nitrosation of Cys152 on activity, it would appear a priori that persulfidation of GAPDH would also be inhibitory. Consistent with this prediction, persulfidation of purified GAPDH at Cys152 was shown to inhibit activity.⁶⁸⁹ A caveat of this study is that persulfidation induced by NaHS or polysulfides was observed at Cys156 and Cys247 in wild-type enzyme but not at Cys152 although modification at these other two sites did not affect enzyme activity. Cys152 was persulfidated only when Cys156 was mutated to serine, and while the mutation did not affect the activity of unmodified enzyme, activity was inhibited upon persulfidation at Cys152.⁶⁸⁹

9.5. Persulfidation of NFkB

Nuclear factor κB (NF-κB) is an antiapoptotic transcription factor. Under basal conditions, it is sequestered in the cytosol via interaction with the inhibitor, IκBα.⁶⁹⁰ During inflammation, cells produce tumor necrosis factor α (TNF-α), which can potentially lead to cell death.⁶⁹¹ H₂S is known to protect against inflammation, but the underlying mechanism is not known.⁶⁹² The discovery that NF-κB is persulfdiated at Cys38 in the p65 subunit suggests a potential mechanism of H₂S action.⁵⁵² Persulfidation was suggested to promote binding of NF-κB to the coactivator ribosomal protein S3, increasing its binding to promoters of antiapoptotic genes, including CSE (Figure 30). An opposing mechanism was however suggested by the report that H₂S suppresses oxidized low-density lipoprotein-induced macrophage inflammation by inhibiting NF-κB.⁶⁹³ Persulfidation of Cys38 was suggested to lead to cytoplasmic retention of NF-κB, inhibiting its DNA binding activity.

Figure 30.

Persulfidation of NF-κB may regulate apoptosis. The proinflammatory cytokine TNFα, involved in the control of inflammatory reactions, stimulates CSE transcription by activating the SP1 transcription factor, resulting in increased H₂S levels. H₂S induces persulfidation of Cys38 in the p65 subunit of NF-κB, enhancing the binding of NF-κB subunits to the coactivator RPS3. The activator complex then migrates to the nucleus where it upregulates the expression of several antiapoptotic genes. TNFα, tumor necrosis factor α; CSE, cystathionine γ lyase; p50 and p65 subunits of NF-κB; RPS3, ribosomal protein S3; SP1, specificity protein-1.

9.6. MEK1/PARP-1 Activation and DNA Damage Repair

DNA damage stimulates a complex and highly concerted DNA damage repair response, which includes binding of poly(ADP- ribose)ation polymerases (PARPs) to DNA strand breaks and catalysis of poly(ADP-ribose)ation. Poly(ADP-ribose)ation attracts other DNA damage repair proteins.⁶⁹⁴ The MEK/ERK signaling cascade plays an important role in activating PARPs.⁶⁹⁵ Persulfidation of Cys341 in MEK1 reportedly facilitates the translocation of phosphorylated ERK1/2 into the nucleus, where it activates PARP-1 and increases the DNA damage repair yield.⁵⁵⁷

9.7. Persulfidation of Parkin

Mutations in parkin, an E3 ubiquitin ligase, are associated with the etiology of Parkinson’s disease, which is caused by the death of dopamine-generating cells in the substantia nigra.^696,697 Parkin contains reactive cysteine residues that are susceptible to oxidative modifications. For example, S-nitrosation of parkin inhibits its activity.⁶⁹⁸ Parkin can be persulfidated at Cys59, Cys95, and Cys182.⁵⁹⁴ The activity of parkin is reportedly increased upon persulfidation and is correlated with the rescue of neurons from cell death by removal of damaged proteins (Figure 31). A decrease in parkin persulfidation has been reported in brain from Parkinson’s disease patients, while S-nitrosation is increased in the same samples.⁵⁹⁴ If substantiated, the activity of parkin would appear to be differentially regulated by modifications at the same cysteine residues and H₂S donors could have therapeutic potential in the early treatment of Parkinson’s disease.

Figure 31.

Possible regulatory role of H₂S on the catalytic activity of parkin. (A) In healthy subjects, parkin, a E3 ubiquitin ligase, is persulfidated, which increases its enzymatic activity. This leads to ubiquitination of diverse substrates and their subsequent proteasomal degradation. (B) In patients with Parkinson’s disease, parkin is S-nitrosated. The decreased catalytic activity results in protein aggregation, accumulation of toxic proteins, and cell death.

Parkin is also a regulator of mitophagy, which leads to the removal of damaged mitochondria, particularly during ischemia-reperfusion injury.⁶⁹⁹ The pharmacological potential of H₂S in preventing ischemia-reperfusion injury,^677,678 might be mediated in part by the persulfidation of parkin.

9.8. Persulfidation of the TRP Channels

CBS-deficient patients exhibit a variety of phenotypes, including osteoporosis,⁷⁰⁰ which is characterized by a low bone density and increased risk of fracture. Bone marrow mesenchymal stem cells are nonhematopoietic multipotent stem cells responsible for bone formation and balancing osteoclast-mediated bone resorption to maintain bone mineral density. H₂S donors reportedly protect MC3T3-E1 osteoblasts against H₂O₂-induced oxidative damage, although the mechanism of this effect is not known.⁷⁰¹ CBS deficiency reportedly resulted in aberrant intracellular Ca²⁺ influx due to reduced persulfidation of multiple TRP channels.⁵⁵⁵ Decreased Ca²⁺ influx downregulates PKC/Erk-mediated Wnt/β-catenin signaling, which is important for controlling osteogenic differentiation of bone marrow mesenchymal stem cells that are postulated to produce H₂S to regulate their self-renewal and osteogenic differentiation.⁵⁵⁵

9.9. Persulfidation Targets Revealed by Proteomic Approaches

Early studies on xanthine oxidase and aldehyde oxidase, which are molybdopterin containing enzymes, reported that they are regulated by persulfidation. However, the location of the persulfide was not established, and based on later structural and spectroscopic studies, the labile sulfur was found to be a sulfide ligand to molybdenum rather than persulfide.⁷⁰² Early studies on Cu,ZnSOD reported the presence of an absorbance peak at 325 nm⁷⁰³ that was assigned to persulfidation at Cys111.⁷⁰⁴ Persulfidation blocked copper-induced protein aggregation but did not affect SOD activity.⁷⁰⁵ In addition to being modified by persulfidation at Cys111, two SOD monomers can be covalently linked via a polysulfane bridge (up to 5 sulfur atoms) between their Cys111 residues.⁷⁰⁶

Persulfide proteome analysis in response to increased endogenous H₂S levels due to ER stress⁵⁵⁸ or exogenous treatment with GYY4137 and polysulfides⁵⁵⁹ has identified many persulfidated proteins. Under ER stress conditions, a total of 827 proteins were identified⁵⁵⁸ of which 178 overlapped with the much smaller 208 persulfidated protein data set reported in the GYY4137 treatment study.⁵⁵⁹ Enrichment of persulfidated proteins involved in translation, glycolysis, and the TCA cycle was reported in both studies in addition to the heat shock proteins Hsp70 and Hsp90 proteins and proteins involved in actin remodeling (actin, actinin, cofilin, and actin-related proteins),^558,559 Actin and Hsp70 and Hsp90 were also identified previously in studies that yielded very limited persulfide target identification.^19,165,511 Significant overlap was observed between persulfidation, sulfenylation, S-nitrosation and glutathionylation targets.^558,559 Although no consensus sequence motif could be identified around persulfidated cysteines, there was some enrichment of the location of target cysteines at the N-termini of alpha helices.⁵⁵⁸

10. PHYSIOLOGICAL AND PHARMACOLOGICAL EFFECTS OF H₂S

The past two decades have witnessed an increasing interest in understanding the physiological and pharmacological effects of H₂S and its donors. However, with the recent development of analytical tools for H₂S detection (see section 3.4), it has become clear that the actual endogenous values of H₂S are quite low (see section 3.5), while the vast majority of physiological experiments were performed with supra-physiological amounts of H₂S. Therefore, in this section, we provide an overview of processes relevant to human biology that are regulated by endogenous H₂S or are responsive to the pharmacological treatment of H₂S or its donors. Microbial^707–713 and plant^714–722 sulfur metabolism and the physiological roles of H₂S in these systems are not covered.

10.1. H₂S and the Nervous System

H₂S affects hippocampal long-term potentiation by acting on N-methyl-D-aspartate (NMDA)-type glutamate receptors albeit only when applied together with weak tetanic stimulation at active but not quiescent synapses.⁶ Under these conditions, H₂S facilitated NMDA receptor-mediated currents by activating adenylyl cyclase and the downstream cyclic adenosine mono-phosphate (cAMP)/protein kinase A (PKA) cascades.⁷²³ Although persulfidation of the NMDA receptor has been implicated, it has not been demonstrated.

Several ion channels have been identified as potential targets of H₂S.^724–736 H2S increases intracellular Ca²⁺ levels and subsequent Ca²⁺ waves in primary astrocyte cultures and hippocampal slices from rats.⁷²⁴ Similar results were obtained with microglial cells and with a neuroblastoma cell line.^725,726 Furthermore, H₂S affected intracellular acidification in a concentration-dependent manner in primary cultured microglia and astrocytes by regulating the activities of the Cl⁻/HCO₃⁻ exchanger and Na⁺/H⁺ exchanger.⁷²⁷

H₂S is postulated to have both pro- and antinociceptive effects in the peripheral nervous system.^728–736 Colonic luminal administration of NaHS caused nociceptive behavior manifested as abdominal allodynia/hyperalgesia in mice, which was abolished by a T-type channel blocker.⁷²⁸ NaHS-induced nociception also caused reversible T-type Ca²⁺ channel-dependent hyperalgesia in the rat spinal cord and peripheral tisssues.^730,731 These results suggest that sensitization/activation of T-type Ca²⁺ channels might be involved^728,731 although TRP channels have also been suggested to mediate pronociceptive effects of H₂S in rodent models.^732–734 In contrast, subcutaneous injection of millimolar H₂S solutions did not cause pain in humans.²⁶⁷ The antinociceptive effects of H₂S have been linked to activation of K_ATP channels.^129,735,737 However, most of these effects were caused at high, rather than pharmacological, doses of H₂S.

As a pharmacological agent, H₂S can improve disease outcomes in different pathological settings (Figure 32). For example, neuronal cell death caused by peroxynitrite was significantly attenuated by H₂S treatment.⁷³⁸ Neuronal injury induced by H₂O₂ in primary cultured rat astrocytes impaired glutamate uptake, whereas treatment with H₂S exerted a neuroprotective effect by increasing glutamate uptake.⁷³⁹ Pharmacological H₂S treatment has also shown some promise for treating Parkinson’s disease as discussed in section 9.7.⁵⁹⁴

Figure 32.

Some possible physiological roles of H₂S in neurodegenerative disorders. H₂S has been shown to be involved in pathogenesis of Huntington’s, Alzheimer’s, and Parkinson’s diseases, spinocerebellar ataxia, and traumatic brain injury. In healthy subjects, the expression of CSE is regulated by SP1 and ATF4 transcription factors. In Huntington’s disease, abnormal mutated huntingtin (mHtt) protein binds to SP1 and inhibits its activity. Reduced CSE expression results in oxidative stress that subsequently affects ATF4 expression. H₂S also inhibits the production of amyloid beta (Aβ) at different catalytic steps of Aβ. The mature isoform of APP is cleaved by β- and γ-secretases forming Aβ. H₂S interferes with APP maturation and inhibits the activity of β- and γ-secretases leading to the decreased production of Aβ. In Parkinson’s disease, H₂S induces persulfidation of parkin and increases E3 ubiquitin ligase activity. H₂S has beneficial effects in spinocerebellar ataxia type 3, where it regulates protein persulfidation and improves SCA3-associated tissue degeneration. In traumatic brain injury H₂S exerts antiapoptotic effects and down-regulates the expression of autophagy-related proteins, reduces brain edema and improves the recovery of motor and cognitive dysfunction.

H₂S levels are reported to be substantially reduced in brain of Alzheimer’s disease patients as compared to healthy individuals.^740,741 In a rat model of Alzheimer’s disease, pretreatment with NaHS improved learning and memory deficits.⁷⁴² Treatment of PC12 cell line with H₂S inhibited expression of BACE-1 (beta-site amyloid precursor protein cleaving enzyme-1) mRNA and protein, a major β-secretase involved in amyloid beta (Aβ) production.⁷⁴³ The PI3K/Akt signaling pathway is reportedly involved in the H₂S-induced decrease in BACE-1 expression and Aβ release. H₂S treatment of SH-SY5Y cells suppressed Aβ formation probably by inhibition of amyloid precursor protein glycosylation and γ-secretase activities.⁷⁴⁴ These studies suggest that the H₂S might exerts its effects on different steps involved in Aβ generation (Figure 32).^743,744

H₂S production is reported to be markedly decreased in Huntington’s disease.^745,746 Mutant huntingtin protein inhibits Sp1, a transcriptional activator of CSE, leading to decreased protein expression and consequently, H₂S production (Figure 32).⁷⁴⁶ H₂S showed beneficial effects for spinocerebellar ataxia type 3 (SCA3), a neurodegenerative disease caused by polyQ repeats in ataxin-3, which leads to protein aggregation and subsequent neuronal dysfunction and death.⁷⁴⁷ In a Drosophila model of SCA3, CSE overexpression or treatment with thiosulfate reduced levels of oxidized proteins, inhibited the immune response and prevented SCA3-associated tissue degeneration. These beneficial effects were correlated with an increase in persulfidation levels.⁷⁴⁷

10.2. H₂S and the Cardiovascular System

H₂S exerts multiple effects in the cardiovascular system including attenuating myocardial ischemia reperfusion injury, promoting angiogenesis, relaxing smooth muscle cells, and regulating blood pressure.^748–752 CSE is believed to be the primary H₂S producing enzyme in the cardiovascular system. It is expressed in vascular endothelial cells, smooth muscle cells, and cardiomyocytes.^12,672,753

Relaxation of rat aortic tissue in vitro was one of the first described effects of H₂S in the cardiovascular system and occurred in synergy with NO^•.⁷ Intravenous application of H₂S decreased blood pressure in rats, which was suppressed by glibenclamide, a K_ATP channel blocker.^{672,754–756} Exposure of isolated vascular smooth muscle cells to H₂S increased K_ATP currents. A vasodilatory role for endogenously synthesized H₂S appeared to be supported by the observation that CSE knockout mice develop age-related hypertension.¹² However, a contradictory result was reported by a second group, which found no changes in blood pressure in CSE knockout mice.⁷⁵⁷ H₂S has been described as an endothelium-derived hyperpolarizing factor (Figure 33),⁷⁵⁶ acting primarily via activation of the K_ATP channel, which can be persulfidated at Cys43 in the pore-forming Kir 6.1 subunit⁵⁵³ (see section 9.1). However, the mechanism by which Cys43 is persulfidated and whether H₂S acts alone as an endothelium-derived hyperpolarizing factor are unclear.

Figure 33.

Possible signaling roles of H₂S in the vascular system. At a sensory nerve ending, H₂S interacts with NO^• to give HNO. HNO activates TRPA1 channels; this results in Ca²⁺ influx and subsequent release of calcitonin gene-related peptide (CGRP). Binding of CGRP to its receptor on vascular smooth muscle cells activates the adenylate cyclase and the cyclic adenosine monophosphate (cAMP)-controlled downstream signaling pathways. As a result of elevated cAMP, protein kinase A (PKA) is activated and could potentially increase the activity of eNOS. Persulfidation of Cys443 on eNOS increases the activity of the enzyme as well as its ability to be phosphorylated, which results in increased production of NO^• and activation of soluble guanylate cyclase (sGC). H₂S potentiates the binding of NO^• to sGC by reducing ferric heme to the ferrous state. Degradation of cGMP by phosphodiesterase (PDE) is prevented by H₂S. These effects result in vasodilation of smooth muscle cells. Persulfidation of Cys43 on K_ATP channel enhances its activity, resulting in the influx of K⁺ and hyperpolarization of vascular smooth muscle cells. H₂S also plays a role in angiogenesis. Binding of VEGF to its receptor on endothelial cells induces the production of H₂O₂ by activating NADPH oxidase (Nox). H₂O₂ supposedly increases CSE expression leading in turn, to increased H₂S production. H₂S stimulates activation of the Akt signaling cascade, which results in the phosphorylation of eNOS, increasing its activity. NO^• acts as a pro-angiogenic factor.

There is growing evidence that the vasodilatory effects of H₂S are intricately linked to NO^• signaling pathways.^{7,266–268,465,758} For instance, the K_ATP channel-based vasodilatory effect of H₂S is attenuated in the absence of NO^•,^672,754 and inhibition of eNOS abrogated H₂S-induced vasorelaxation.^7,266,267 Furthermore, HNO, a product of the reaction between H₂S and NO^•, activates TRPA1.²⁶⁷ Stimulation of the calcitonin gene-related receptor on smooth muscle cells activates adenylate cyclase, which then generates cAMP, a powerful secondary messenger responsible for vasodilation (Figure 33).²⁶⁷

Regulation of cGMP levels appears to be an important mechanism by which H₂S potentiates the effects of NO^• particularly in the context of angiogenesis.^{266,410,758,759} H₂S inhibits the phosphodiesterase^410,758,759 slowing cGMP degradation and increasing its half-life. Furthermore, binding of H₂S to soluble guanylate cyclase leads to reduction of the heme iron to the ferrous state, which binds NO^•377 (Figure 33). H₂S also regulates eNOS activity and expression. Phosphorylation of eNOS increases its enzymatic activity and is enhanced in cells and blood vessels treated with H₂S apparently via persulfidation at Cys443^266,268,454 (Figure 33). Persulfidation increases eNOS activity and its ability to be phosphorylated, while also preventing the inhibitory S-nitrosation modification at the same cysteine residue.⁴⁵⁴ Surprisingly, H₂S inhibits neuronal NOS and inducible NOS by directly binding to the iron heme center but does not inhibit purified eNOS.

Circulating H₂S levels in patients with chronic heart disease or heart failure are significantly reduced compared to age-matched controls.⁷⁶⁰ The pharmacological potential of H₂S has been demonstrated in a myocardial ischemia/reperfusion injury model^677,678,750 where H₂S administration at the time of reperfusion reduced infarct size by 72%.⁷⁵⁰ The protective effect of H₂S was linked to preservation of mitochondrial function, reduction of cardiomyocyte apoptosis, anti-inflammatory responses, and antioxidant effects.⁷⁵⁰ Heart Nrf-2 and phosphorylated Akt levels were significantly higher in H₂S treated mice suggesting that antioxidant gene expression was increased.⁶⁷⁷ Upregulation of the antiapoptotic Blc-2 protein and down-regulation of the pro-apoptotic factors Bax and caspase 2 were seen after H₂S treatment.⁷⁶¹ These effects could be regulated by persulfidation of Keap1 as described in section 9.2.

Binding of vascular endothelial growth factor (VEGF) to its receptor increases CSE expression via intermediate production of H₂O₂ as a signaling molecule.⁷⁶² H₂S production stimulates the Akt pathway, resulting in eNOS phosphorylation and higher NO^• levels. The VEGF receptor might be a direct target of H₂S, which reportedly reduces the Cys1045−Cys1024 disulfide bond and disrupts the active conformation of the receptor.⁷⁶³

Antiatherosclerotic properties of H₂S have been reported.^764,765 CSE knockout mice fed an atherogenic diet developed atherosclerotic lesions and exhibited a different plasma lipid profile than wild-type mice.⁷⁶⁵ Expression of the adhesion molecule ICAM-1, which is important for the development of atherosclerotic lesions, was significantly elevated in aorta of CSE knockout mice on an atherogenic diet.⁷⁶⁴ Additionally, NFκB expression was elevated in smooth muscle cells isolated from CSE knockout mice.⁷⁶⁴

10.3. H₂S and Inflammation

H₂S reportedly elicits proinflammatory and anti-inflammatory effects in various models of inflammation.⁷⁶⁶ H₂S had a proinflammatory effect on acute pancreatitis and associated lung injury and treatment with the CSE inhibitor, D,L-propargylglycine, significantly decreased pancreatic and lung injury.⁷⁶⁷ The proinflammatory effect of H₂S was diminished in CSE knockout mice in which acute pancreatitis and associated lung injury was induced.⁷⁶⁸ Furthermore, in an ischemia reperfusion injury model, a reduced inflammatory response was observed in kidneys of CSE knockout mice.⁷⁶⁹ A proinflammatory response to H₂S has also been reported in several models of sepsis,^770–772 and H₂S levels are increased in patients with septic shock.⁶⁹²

An anti-inflammatory effect of H₂S has been reported in subjects with intestinal ischemic damage and ethanol-induced gastritis.^773–775 CSE expression is decreased in the gastric mucosa by nonsteroidal anti-inflammatory drugs^776–778 and NaHS treatment decreased expression of TNF-α, intercellular adhesion molecule 1 (ICAM-1), and lymphocyte-associated antigen-1.^776,777 H₂S synthesis is markedly increased in colon ulcers, and it promotes the rapid restoration of the epithelial barrier integrity and the repair of the damaged tissue.⁷⁷⁹ The slow releasing H₂S donor GYY4137 also exerts an anti-inflammatory effect^121,780 by inhibiting NFkB activity. Persulfidation of the p65 subunit of NFkB is proposed to increase its interaction with the ribosomal protein S3 and to upregulate several antiapoptotic genes (see section 9.5).⁵⁵²

10.4. H₂S and the Respiratory System

H₂S treatment reportedly attenuates bleomycin-induced pulmonary fibrosis in rats.⁷⁸¹ H₂S treatment suppressed the migration, proliferation, and myofibroblast trans-differentiation of a human lung fibroblast cell line. The inhibitory effects of H₂S were correlated with a decrease in ERK phosphorylation.^782,783 Transforming growth factor β1 (TGF-β1) is a master regulator of fibrosis, and its inhibition by H₂S resulted in decreased vimentin expression and increased E-cadherin levels.⁷⁸⁴ A bronchodilatory effect of H₂S was ascribed to inhibition of Ca²⁺ release from the ER.⁷⁸⁵ H₂S treatment induced relaxation of mouse tracheal smooth muscle cells by activating the calcium-activated potassium channel.⁷⁸⁶ Furthermore, H₂S reportedly plays a role in the pathology and treatment of chronic obstructive pulmonary disease.⁷⁸⁷

10.5. H₂S and the Renal System

H₂S appears to play an important role in the onset and progression of renal diseases. Plasma H₂S levels are lower in patients with diabetic versus nondiabetic nephropathy undergoing chronic hemodialysis.⁷⁸⁸ CBS and CSE expression is downregulated in experimental models of diabetes.^789,790

An intrarenal infusion of NaHS increased blood flow, glomerular filtration rate, and urinary sodium and potassium excretion in rats.^791,792 H₂S inhibited the Na⁺/K⁺/2Cl⁻ cotransporter and Na⁺/K⁺/ATPase activity in proximal kidney tubule epithelial cells⁷⁹² and decreased cAMP levels in different renal cell types.^793,794 H₂S also decreased renin production in rat kidney.⁷⁹⁴ CSE overexpression increased endogenous H₂S production and suppressed isoproterenol-induced renin release.⁷⁹⁴ However, the underlying mechanism by which H₂S regulates renin release is unclear.

Conflicting results on the effects of H₂S on kidney ischemia/reperfusion injury have been reported.^769,795 Increased damage and mortality after renal ischemia/reperfusion was reported in CSE knockout mice, and H₂S treatment protected mice from ischemia-induced renal injury and decreased mortality.⁷⁹⁵ A second study using a different strain of CSE knockout mice failed to observe a significant difference between wild-type and knockout animals exposed to kidney ischemia/reperfusion.⁷⁶⁹

10.6. H₂S and the Liver

All H₂S-producing enzymes are expressed in the liver, which plays an important role in glucose and lipid homeostasis, xenobiotic metabolism, and antioxidant defense.^796–799 H₂S is suggested to be an important modulator of hepatic micro-circulation.⁸⁰⁰ H₂S donors such as diallyl trisulfide attenuate ethanol-induced liver injury in mice and increase the activity of mitochondrial antioxidant enzymes.⁸⁰¹ The hepatoprotective effects of H₂S donors were correlated with Nrf2 translocation and increased expression of antioxidant genes, which is regulated by Keap1 persulfidation (see section 9.2).^554,679

10.7. H₂S and the Gut

Some gut microbes use sulfate as a terminal electron acceptor for respiration and produce H₂S using the dissimilatory sulfite reductase enzyme complex.^710,712 Desulfovibrio is the predominant sulfate reducing bacterium in the human intestine, while Desulfobacter, Desulfomonas, Desulfobulbus, and Desulfotomaculum are found at lower levels.⁸⁰² Very high H₂S (up to 1000 ppm) has been reported in the rat cecum,⁸⁰³ and 0.2−30 ppm of H₂S has been reported in human flatus.⁸⁰⁴ It is estimated that approximately half of the fecal H₂S is produced by gut microbes with the remainder being derived from host metabolism.⁸⁰⁵ An imbalance in the number or composition of gut microbes is associated with various diseases.⁷¹² Sulfate reducing bacteria are resistant to a broad spectrum of antibiotics, and repeated use of these drugs might favor a bloom of these bacteria.⁸⁰⁶ An increase in the number of sulfate reducers has been observed in patients with ulcerative colitis, inflammatory bowel disease and Crohn’s disease,⁸⁰⁷ periodontitis,⁸⁰⁸ pouchitis,⁸⁰⁴ and irritable bowel syndrome.⁸⁰⁹ Probiotic and prebiotic treatments reduce the numbers of Desulfovibrio bacteria^810,811 and the levels of proinflammatory cytokines IFN-γ, TNF-α, and IL-1β.⁸¹¹

10.8. H₂S and the Reproductive System

CBS and CSE are expressed in the human corpus cavernosum, the muscular trabeculae, and smooth muscle components of the penile artery.^812,813 The administration of CSE inhibitors into corpus cavernosum impaired the normal intracavernosal pressure response to cavernous nerve electrostimulation suggesting a possible role for H₂S in penile tissue smooth muscle relaxation.^814,815 NaHS treatment induced relaxation of rabbit⁸¹⁵ and human⁸¹² corpus cavernosum strips in a concentration-dependent manner. An H₂S-donating derivative of sildenafil has been developed for potential use in treating erectile dysfunction.⁸¹⁶

H₂S also exerts effects on male and female fertility.^817–823 Expression of H₂S-producing enzymes and H₂S synthesis have been reported in uterus, vagina, and placenta.^818–821 NaHS was shown to reversibly attenuate the contractile response of isolated rat uterus and delay parturition^824–827 and mediate spontaneous contractions of the human oviduct.⁸²³

10.9. H₂S and Oxygen Sensing and Hibernation

Exposure to low concentrations of H₂S (80 ppm) induced a suspended animation-like state in mice, by decreasing metabolic rate and core body temperature.¹³ The H₂S-induced depression of the metabolic rate observed in mice could be beneficial in patients with major trauma or cardiac arrest.⁸²⁸

Hypoxia increases H₂S production in carotid bodies of rat and mice.⁸²⁹ In CSE knockout mice and wild-type rats treated with a CSE inhibitor, the hypoxic sensitivity of carotid bodies is markedly impaired. CO reportedly abolished CSE activity and reduced H₂S generation in rat carotid bodies via protein kinase G-dependent phosphorylation of CSE.^829,830 In hypoxia, CO levels and CSE phosphorylation decrease, leading to increased H₂S production (Figure 34).²⁶⁰

Figure 34.

Possible H₂S effects on glomus cells of carotid bodies under hypoxic conditions. Under hypoxic conditions the levels of H₂S in glomus cells of carotid bodies are elevated. This could be a result of decreased phosphorylation of CSE due to the lack of CO produced by heme oxygenase-2 (HO-2). The lack of CO results in the inhibition of cyclic guanosine monophosphate (cGMP)-stimulated activation of phosphokinase G. The overall increase in H₂S levels activates the L-type Ca²⁺ and T-type voltage-gated Ca²⁺ channels and mobilizes Ca²⁺ from the endoplasmic reticulum (ER).

The stimulatory effect of H₂S on carotid body sensory activity is completely abolished by cadmium chloride, a nonselective inhibitor of voltage-activated Ca²⁺ channels.^829,831 H₂S treatment and hypoxia induced an increase in intracellular Ca²⁺ concentrations in rat glomus cells. Inhibition of H₂S production prevented carotid body activation and hypertension in rodents exposed to intermittent hypoxia, which is a model for obstructive and central sleep apnea.

Treatment of Caenorhabditis elegans with H₂S increased thermotolerance and longevity at higher temperatures.⁸³² Several proteins were associated with these effects including hypoxia-inducible factor-1 and SKN-1, a homologue of mammalian Nrf-2. Furthermore, knockout of the MST ortholog reduced lifespan in C. elegans,⁸³³ and the effect was rescued by GYY4137. H₂S and ROS are postulated to play important roles in extending lifespan in C. elegans.¹⁵⁶ Decreased expression of CBS significantly reduced the lifespan of germline-deficient C. elegans mutants compared to the wild type strain.¹⁵⁶ The role of protein persulfidation in lifespan extension has not been examined.

10.10. H₂S and Cancer

The roles of H₂S in cancer development and progression are still controversial. Some of the biological effects of H₂S that might be relevant to cancer biology include stimulation of angiogenesis, regulation of intracellular signaling and cell death, and cellular bioenergetics.^834–836 Many of the effects exhibit a biphasic dose−response curve; low concentrations of H₂S are cytoprotective and high concentrations are cytotoxic.^835–839

Overexpression of CBS at protein and mRNA levels has been reported in primary epithelial ovarian cancer tissue^840–842 and in several breast cancer cell lines.^843–845 However, CBS expression is reportedly suppressed in gastric and colorectal cancers, glioma, and hepatocellular carcinoma.^846–848 Additional studies are needed to elucidate the roles of CBS in cancer development and progression. Similar contradictory observations have been reported for CSE, which was shown to be both up- and down-regulated in several cancer types.^849–853 The pro-carcinogenic effects of H₂S also contradict studies using H₂S donors, which show anticarcinogenic effects.^{837–839,854} A deeper understanding of whether and how H₂S plays a role in cancer etiology and progression is needed.

11. CONCLUSIONS

While an increasing number of signaling roles and physiological effects are being attributed to H₂S, our understanding of the underlying mechanisms lags far behind. Significant barriers that have thwarted progress in the field include the technical challenges of working with a redox active and volatile molecule whose salts are often contaminated with polysulfides, which can be more reactive than H₂S itself. The scarcity of sensitive, rigorous, and readily available methods for quantifying H₂S or persulfides poses additional challenges. The identities of the preferred targets for H₂S and of its downstream signaling intermediates are not yet completely understood. While persulfide proteomic analyses are beginning to reveal a rich trove of protein targets, the functional validation of how persulfidation affects individual proteins and metabolic flux has been challenging, since quantitative methods for introducing the persulfidation modification and stabilizing it under assay conditions are not readily available. Solving this methodological problem would lead to insights into how H₂S signals and could identify important therapeutic targets. It would also lead to the resolution of the many contradictory effects of H₂S that have been reported with purified proteins, at the cellular and organismal levels.

The contributions of the enzymes in H₂S-producing and H₂S-oxidizing pathways in controlling H₂S levels and the conditions that permit spiking of H₂S from low nanomolar steady-state concentrations to levels which trigger signaling are other important gaps in the field. Furthermore, the relative contributions of the three H₂S-producing enzymes in different tissues are largely unknown. The development of selective inhibitors for CBS, CSE, MST, and SQR in particular will be very useful as molecular reagents together with gene editing technology for dissecting the roles of these enzymes in H₂S homeostasis. Expanding the tool set for selective and ratiometric fluorescence visualization of protein persulfidation will facilitate elucidation of the spatiotemporal changes that occur during H₂S signaling.

The field of H₂S biology would benefit greatly by being strongly grounded in chemical studies on H₂S and persulfide formation and decay kinetics and reactivity, to provide a more rigorours framework for understanding the potential roles of persulfidation in cell signaling. Proteomic studies suggest significant overlap between persulfidation and other oxidative cysteine modifications. Teasing apart why modifications like S-nitrosation, glutathionylation, and sulfenylation or persulfidation target the same cysteines in some proteins will provide important insights into cross-talk between oxidative cysteine-based signaling pathways.

Regulation of persulfidation is another area that is poorly understood. Are catalysts involved, and if so, what are their identities? Enzyme-catalyzed trans-persulfidation reactions are expected to be important both for adding and removing the persulfide modification, and an understanding of how each process is regulated and how target specificity is achieved is critical for elucidating H₂S-based signaling. The relative newness of the H₂S chemical biology field and the large fundamental gaps in our understanding combine to portend an exciting future.

Biographies

Milos R. Filipovic received his undergraduate and postgraduate training in biochemistry (University of Belgrade). After his postdoc at Friedrich-Alexander University Erlangen-Nuremberg, where he also worked as a group leader for a few years, he moved to Institut de Biochimie et Génétique Cellulaires (IBGC, CNRS UMR5095) at the University of Bordeaux, as Idex Junior Chair and ATIP-Avenir awardee to lead the “Signaling by gasotransmitters” group. He is interested in redox signaling, in general, and in developing new tools to dissect molecular mechanisms by which different gasotransmitters signal in the human body, in particular.

Jasmina Zivanovic studied biology at the University of Belgrade where she also obtained her Ph.D. (in endocrinology). She is a postdoc researcher at IBGC CNRS UMR5095 in the “Signaling by gasotransmitters” group. Her research focus is on developing new tools for persulfide labelling to address the changes of persulfidation in ageing and ageing-related diseases.

Beatriz Alvarez is Associate Professor of Enzymology at the School of Sciences, University of the Republic, Montevideo, Uruguay. She received her M.Sc. and Ph.D. in Chemistry degrees from the University of the Republic. Her interests span the areas of redox biochemistry, kinetics and enzymology, especially in relation to biological thiols and hydrogen sulfide.

Ruma Banerjee was raised as a peripatetic army brat changing ten schools before graduating. She received her formal training in plant science (B.S. and M.S., University of Delhi), biochemistry (Ph.D., Rensselaer Polytechnic Institute, NY), and biophysics (postdoc, University of Michigan). She is currently the Vincent Massey Collegiate Professor and Associate Chair of Biological Chemistry at the University of Michigan Medical School. Her interests range from B₁₂ trafficking and metalloenzymology to hydrogen sulfide homeo-stasis and signaling. She is an Associate Editor for Chemical Reviews and for the Journal of Biological Chemistry.