Enzymology of Hydrogen Sulfide Turnover

Abstract

Hydrogen sulfide is a biological signaling molecule that is produced by organisms ranging from bacteria to man. Since it is also toxic at high concentrations, strategies exist for its efficient removal and consequent maintenance of low steady-state levels in mammals. Enzymes in the sulfur metabolic network responsible for H₂S biogenesis include the cytoplasmic cystathionine β-synthase and γ-cystathionase enzymes of the transsulfuration pathway as well mercaptopyruvate sulfurtransferase that is both mitochondrial and cytoplasmic. The precursors for H₂S are the amino acids cysteine and homocysteine and a cysteine derivative i.e. mercaptopyruvate. H₂S is cleared via a mitochondrial sulfide oxidation pathway that begins with sulfide quinone oxidoreductase and includes a persulfide dioxygenase, rhodanese and sulfite oxidase. The major oxidation products of H₂S are thiosulfate and sulfate. This chapter focuses on the structural enzymology of H₂S biogenesis and oxidation, emphasizing recent advances in the field.

Nicole Motl and Pramod Yadav are equal contributors.

This work was supported in part by the National Institutes of Health (HL58984).

1.1 Introduction to Sulfide Metabolism

Long known as a toxic gas, sulfide like cyanide, targets cellular respiration by reversible inhibition of cytochrome c oxidase. Cells therefore evolved strategies for handling the twin challenges of averting toxicity problems while exploiting the signaling potential of hydrogen sulfide (H₂S), which elicits profound physiological effects (Kimura 2010). Hence, a “safe” window must exist within which low intracellular concentrations of H₂S are maintained and allowed to transiently spike to allow passage of a signal. The concentration width of this window could span three orders of magnitude since steady-state intracellular concentrations of H₂S are estimated to be in the 15–30 nM range (Furne et al. 2008; Levitt et al. 2011; Vitvitsky et al. 2012) while mammalian mitochondrial ATP production is abolished at 50 μM H₂S (Bouillaud and Blachier 2011). The steady-state concentrations of H₂S is a product of both the metabolic flux through the sulfide biogenesis and sulfide oxidation pathways (Vitvitsky et al. 2012) (Fig. 1.1).

Fig. 1.1

Scheme showing pathways of biogenesis and catabolism of H₂S. DAO is D-amino acid oxidase

Three enzymes in the mammalian sulfur metabolic network catalyze H₂S biogenesis (Kabil and Banerjee 2010; Singh and Banerjee 2011). Two of these enzymes reside in the cytosolic transsulfuration pathway and are cystathionine β-synthase (CBS) and γ-cystathionase (CSE). Both are versatile and catalyze H₂S production in a variety of reactions starting from cysteine and/or homocysteine (Chen et al. 2004; Chiku et al. 2009; Singh et al. 2009). The third enzyme, mercaptopyruvate sulfur transferase (MST) (Nagahara et al. 1998; Shibuya et al. 2009), is both cytosolic and mitochondrial in location and transfers the sulfur atom from mercaptopyruvate to an acceptor from where it can be subsequently released as H₂S. MST works in conjunction with the PLP-dependent enzyme, aspartate/cysteine aminotransferase (CAT), which catalyzes the conversion of cysteine to 3-mercaptopyruvate. Recently, D-amino acid oxidase has been reported to convert D-cysteine to 3-mercaptopyruvate, providing an alternative supply route for the MST substrate (Shibuya et al. 2013).

Sulfide oxidation occurs in the mitochondria and connects sulfur metabolism to the electron transfer chain and thereby, to both ATP and reactive oxygen species production (Fig.1.1). Half maximal inhibition of cytochrome c oxidase in cell extracts occurs at concentrations of ~0.3 μM versus ~20 μM H₂S needed to inhibit cellular respiration in intact cells (Bouillaud and Blachier 2011). To avoid intracellular sulfide build-up and consequent toxicity, the sulfide oxidation pathway is activated at considerably lower (~10–20 nM) sulfide concentrations (Bouillaud and Blachier 2011). The high sensitivity of the sulfide oxidation pathway to H₂S, suggests that the duration of H₂S-based signaling is likely to be short (Bouillaud and Blachier 2011). Sulfate and thiosulfate are the major products of the mammalian mitochondrial sulfide oxidation pathway, which comprises four enzymes: sulfide quinone oxidoreductase (SQR), a persulfide dioxygenase (ETHE1) that is the product of the ethe1 gene, rhodanese, and sulfite oxidase (Hildebrandt and Grieshaber 2008). This chapter focuses on the structural enzymology and regulation of H₂S metabolism.

1.1.1 Enzymology of H₂S Biogenesis

1.1.1.1 Cystathionine β-Synthase

Structural Organization of CBS. CBS is a multidomainal protein, which in its role in the transsulfuration pathway, catalyzes the condensation of homocysteine and serine to produce cystathionine (Banerjee et al. 2003). Homocysteine is a redox-active nonprotein amino acid that is produced by hydrolysis of S-adenosylhomocysteine, a product of S-adenosylmethionine (AdoMet)-dependent methylation reactions. Homocysteine is either recycled in the methionine cycle via the action of methionine synthase or committed to cysteine synthesis by the action of CBS (Fig. 1.1). Mutations in CBS are the most common cause of homocystinuria, an autosomal recessive disorder, characterized by severely elevated plasma homocysteine levels (Kraus et al. 1999). CBS deficiency affects multiple organ systems including the ocular, skeletal, cardiovascular and the central nervous system (Mudd et al. 1964).

In addition to the canonical transsulfuration reaction, CBS catalyzes H₂S-producing reactions in which serine is substituted by cysteine (Chen et al. 2004; Singh et al. 2009). These reactions involve β-replacement of cysteine by homocysteine, cysteine or water to generate cystathionine, lanthionine or serine, respectively in addition to the common product, H₂S.

CBS is a homodimeric enzyme. The full-length human enzyme is prone to aggregation and exists in multiple oligomeric states ranging from 2- to 16-mers (Sen and Banerjee 2007). The predominance of the 4-mer in the aggregated mixture has led to confusion in the literature about CBS being a homotetramer. Each monomer is organized into an N-terminal heme-binding domain, a central PLP-binding catalytic core and a C-terminal AdoMet-binding regulatory domain, which contains a tandem repeat of CBS domains. The latter, named after this protein, is found in all three domains of life and refers to a $\upbeta -\upalpha -\upbeta -\upbeta -\upalpha$ secondary structure motif often associated with energy sensing that binds ATP or AMP (Bateman 1997). In CBS, it binds the allosteric activator, AdoMet, which renders transsulfuration flux sensitive to cellular methyl donor status. Thus, when AdoMet levels are low, sulfur is spared and utilized via the methionine cycle to support AdoMet synthesis. In contrast, when AdoMet levels are plentiful, sulfur metabolism is directed towards cysteine synthesis via activation of CBS.

The crystal structures of full-length CBS from Drosophila melanogaster (dCBS, Fig. 1.2a) and a truncated variant of human CBS (hCBS) lacking the C-terminal regulatory domain have been reported (Meier et al. 2001; Taoka et al. 2002; Koutmos et al. 2010). The N terminal heme domain spans residues 1–70 in hCBS and 1–45 in dCBS while the middle PLP domain spans residues 71–411 in hCBS and 46–350 in dCBS. A 27 residue-long linker (351–377) visible in the structure of dCBS connects the PLP and C-terminal domains, which based on sequence alignment, is predicted to extend between residues 382–411 in hCBS.

Fig. 1.2

Structure and catalytic mechanism of CBS. (a) Structure of full-length dCBS. The protein comprises an N-terminal heme-binding domain (red), a PLP domain (blue) and a C-terminal AdoMet-binding domain (yellow). A linker (grey) connects the PLP- and C-terminal domains. (b) Close-up of the active site of hCBS. Hydrogen bonds between PLP and amino acids lining the active site are shown as dotted lines. Figure 1.2a, b were generated using PDB files 3PC2 and 1 M54, respectively. (c) Proposed catalytic mechanism for H₂S generation from cysteine by CBS. The resting enzyme exists as an internal aldimine, which reacts with cysteine to form the external aldimine of cysteine. Elimination of H₂S leads to formation of an aminoacrylate intermediate, which reacts with homocysteine (cysteine or water) to form the external aldimine of the product cystathionine (or lanthionine or serine). The catalytic cycle is completed upon release of product and reformation of internal aldimine

Catalytic Mechanism of CBS. The active site of CBS is lined with residues that position the substrates and cofactor for catalysis. PLP is covalently linked to an active site lysine (K119 in hCBS and K88 in dCBS) and its phosphate moiety is enveloped by a glycine-rich loop (T₂₆₀-G₂₅₉-G₂₅₈-T₂₅₇-G₂₅₆, hCBS numbering) (Fig. 1.2b). N149 and S349 form hydrogen bonds with the exocyclic oxygen and pyridine nitrogen, respectively (Meier et al. 2001; Taoka et al. 2002). Replacement of the corresponding serine in yeast CBS (yCBS) by alanine results in complete loss of activity, while replacement with aspartate results in an ~80-fold lower activity despite the mutants having only two-fold lower PLP occupancy (Quazi and Aitken 2009). Fluorescence and resonance Raman studies demonstrate that the PLP exists in two tautomeric forms: an active ketoenamine and an inactive enolimine (Singh et al. 2009; Weeks et al. 2009; Yadav et al. 2012). N149 helps stabilize the active ketoenamine tautomer. Perturbations in the hydrogen-bonding network in the active site can influence the tautomeric equilibrium, and as described below, is one mechanism by which the heme cofactor exerts its effects on the active site (Singh et al. 2009; Yadav et al. 2012).

The first step in the catalytic cycle (Fig. 1.2c) is binding of serine or cysteine (Fig. 1.2c) to the active site followed by displacement of the lysine residue that forms an internal aldimine with PLP (Banerjee et al. 2003). Abstraction of the α-proton from the resulting external aldimine of serine/cysteine generates a carbanion intermediate. In the crystal structure of dCBS obtained at 1.7 Å resolution, the carbanion intermediate with serine was captured and revealed how the active site stabilizes this reactive species (Koutmos et al. 2010). The Cα atom is clearly sp ² hybridized in this intermediate and the ε-amino group of K88 is within 2.1, 2.5 and 3.0 Å of the Cα, imino nitrogen of the Schiff base and C4A of PLP, respectively, where the negative charge is predominantly localized. Elimination of H₂O or H₂S from the external aldimine of serine or cysteine, respectively leads to formation of the aminoacrylate intermediate, which was also captured in a crystal structure of dCBS obtained at 1.55 Å resolution (Koutmos et al. 2010). In this structure, the Cα is also clearly sp ² hybridized and the ε-amino group of K88, which is swung away from Cα, is parked near the oxygen atoms of the phosphate moiety of PLP. The Cβ of the aminoacrylate intermediate is positioned for nucleophilic attack by the thiolate of homocysteine (to generate cystathionine), or cysteine (to generate lanthionine) or by water (to generate serine). The final step preceding product release involves a second transchiffization reaction in which the active site lysine displaces the product and the enzyme returns to its resting internal aldimine state.

Kinetics of H ₂ S Generation by CBS . Detailed steady-state kinetic analyses of H₂S generation by human and yeast CBS have been reported (Singh et al. 2009). In addition to the canonical β-replacement reaction in the transsulfuration pathway (Eq. 1.1), CBS catalyzes at least three other reactions with a combination of cysteine and homocysteine as substrates, leading to H₂S production (Eqs. 1.2, 1.3, and 1.4). The CBS active site has two pockets for binding amino acids that

$$ \mathrm{ Serine} + \mathrm{ Homocysteine}\to \mathrm{ Cystathionine} + {{\mathrm{ H}}_2}\mathrm{ O} $$

(1.1)

$$ \mathrm{ Cysteine} + \mathrm{ Homocysteine}\to \mathrm{ Cystathionine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.2)

$$ \mathrm{ Cysteine}\to \mathrm{ Serine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.3)

$$ \mathrm{ Cysteine} + \mathrm{ Cysteine}\to \mathrm{ Lanthionine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.4)

are designated as sites 1 and 2. Site 1 binds the amino acid that forms the external aldimine with PLP while site 2, binds the nucleophilic amino acid (Singh et al. 2009). At saturating substrate concentrations, the specific activity for the β-replacement of cysteine by homocysteine (reaction 2) is ~4-fold higher than for the canonical β-replacement of serine by homocysteine (reaction 1). H₂S generation from one (reaction 3) or two (reaction 4) moles of cysteine results in serine or lanthionine and are ~40- and ~20-fold lower, respectively than H₂S generation from cysteine and homocysteine (Singh et al. 2009).

Within the cell, the efficiency of each CBS-catalyzed reaction is dictated in part by the relative concentrations of the individual substrates and the K_M values, which are high for both yeast and human CBS relative to the intracellular concentrations of their amino acid substrates. The K_M for cysteine for hCBS at site 1 is 6.8 ± 1.7 mM and higher still for site 2 (27.3 ± 3.7 mM). For yCBS, the K_M values for cysteine at sites 1 and 2 are 3.6 ± 1.7 mM and 33 ± 3.7 mM, respectively. Homocysteine only binds to site 2 and the K_Ms for yeast and human CBS are 0.13 ± 0.02 mM and 3.2 ± 1.3 mM, respectively (Singh et al. 2009). Simulations were performed using the steady-state kinetic data for hCBS at physiological concentrations of substrate (560 μM serine, 100 μM cysteine and 10 μM homocysteine). The simulations predicted that ~96 % of H₂S derived from hCBS is via β-replacement of cysteine by homocysteine with the remainder being contributed by the other two reactions (Singh et al. 2009).

Pre-steady state kinetic analysis and characterization of reaction intermediates was first reported for yCBS using stopped flow spectroscopy (Jhee et al. 2001; Taoka and Banerjee 2002; Singh et al. 2011). The heme in hCBS interferes with enzyme-monitored pre-steady state kinetic analysis by masking the PLP absorbance. Kinetic analysis of a heme-less hCBS variant lacking the N-terminal heme domain was limited by its poor stability (Evande et al. 2004). We have recently employed difference UV-visible stopped-flow spectroscopy to characterize intermediates in the hCBS-catalyzed reaction (Yadav and Banerjee 2012), an approach, that we used previously to demonstrate the intermediacy of an aminoacrylate intermediate in dCBS (Koutmos et al. 2010). The rate of aminoacrylate formation is ~2.5-fold faster with serine than with cysteine in the reaction catalyzed by hCBS and product release appears to limit the overall reaction rate (Yadav and Banerjee 2012).

Allosteric Regulation of CBS by Heme. The N-terminal heme domain in CBS is devoid of any secondary structure. The heme is hexa-coordinate and its histidine and cysteine axial ligands were predicted by EPR, extended X-ray absorption fine structure (Ojha et al. 2000) and resonance Raman (Green et al. 2001) spectroscopy. The UV-visible spectrum of ferric CBS exhibits a Soret peak at 428 nm and a broad α/β absorption band centered at 550 nm. In the ferrous state, the Soret peak shifts to 449 nm with concomitant sharpening of the α and β absorption bands at 571 and 540 nm (Taoka et al. 1998). Mutation of either heme ligand diminishes hCBS activity (by ~9-fold) despite full PLP saturation in the C52A and C52S mutants and 75 % saturation in the H65R mutant (Ojha et al. 2002). On the other hand, heme saturation is greatly reduced in the heme ligand mutants (19 % and 40 % in the cysteine and histidine ligand mutants, respectively) compared to wild-type CBS. The C52S or C52A mutants have five-coordinate high-spin heme with a Soret peak blue-shifted from 428 to 415–417 nm in the ferric form and from 449 to 423 nm in the ferrous form (Ojha et al. 2002). In the ferrous-CO state, the Soret peaks of the mutants (at ~422 nm) are virtually identical to that of wild-type enzyme, consistent with the substitution of the cysteine ligand by CO. The Soret peaks in the ferric, ferrous and ferrous-CO states in the H65R mutant are at 424, 421 and 420 nm, respectively, consistent with the presence of a low-spin heme (Ojha et al. 2002).

While the role of the heme domain in CBS has been debated, there is growing evidence that it regulates CBS activity in response to changes in the heme spin-or ligation-state (Taoka and Banerjee 2001; Taoka et al. 2001; Singh et al. 2007). The ferrous heme ligands CO and NO, inhibit CBS activity (Taoka et al. 1999; Taoka and Banerjee 2001). Ferrous-NO CBS is five-coordinate with a broad Soret peak at ~390 nm (Taoka and Banerjee 2001). CBS exhibits nonequivalent binding sites for CO with K_d values of 1.5 ± 0.1 μM and 68 ±14 μM for full-length hCBS and 3.9 ± 2 μM and 50 ± 8 μM for the C-terminal truncated form (Taoka et al. 1999; Puranik et al. 2006). Binding of NO to full-length CBS exhibits a K_d of 30 ± 5 μM for NO (Gherasim et al., unpublished results). Binding of CO to full-length CBS inhibits enzyme activity with a K _i value of 5.6 ± 1.9 μM (Taoka et al. 1999). The redox potential of the heme is −350 ± 4 mV (Singh et al. 2009) and −287 ± 2 mV (Carballal et al. 2008) in full-length and truncated hCBS, respectively. This difference between the full-length and truncated forms suggests that the regulatory C-terminal domain modulates the heme redox potential. We have recently demonstrated that despite the low reduction potential of the CBS heme, reversible regulation by CO can be achieved with physiologically relevant reductants (Kabil et al. 2011). The diflavin oxidoreductases, human methionine synthase reductase and novel reductase 1 reduce ferric CBS and the ferrous-CO species is formed in the presence of NADPH and CO.

Communication between the heme and PLP domains occurs via an α-helix whose N-terminal end leads in from the glycine rich loop harboring the conserved T257 and T260 residues that make contact with the phosphate moiety of PLP. At the C-terminal end of the helix, R266 is involved in a salt-bridge interaction with the heme ligand, C52. Changes in the heme coordination state e.g. by formation of the ferrous-CO species, is predicted to disrupt the salt bridge between C52 and R266, displacing the α-helix, which in turn is propagated to the PLP site shifting the tautomeric equilibrium towards the enolimine (Weeks et al. 2009). Interestingly, replacement of R266 by methionine, a mutation described in homocystinuric patients, also results in the predominance of the inactive enolimine. Similarly, substitutions at T257 and T260 in the PLP domain, stabilize the inactive enolimine tautomer leading to significant loss of CBS activity (Yadav et al. 2012). Interestingly, these CBS mutations impact the H₂S and cystathionine-producing versus H₂O and cystathionine-producing reactions unequally, suggesting that that these two activities can be differentially regulated (Yadav et al. 2012).

Allosteric Regulation of CBS by AdoMet. AdoMet binds to the C-terminal regulatory domain (Scott et al. 2004) and enhances CBS activity ~2–3-fold (Finkelstein et al. 1975). Truncation of the C-terminal domain in hCBS leads to loss of AdoMet-dependent regulation, increases CBS activity and also decreases its propensity for aggregation (Kery et al. 1998). Hence, the regulatory domain exerts an autoinhibitory effect that is alleviated upon AdoMet binding, activating mutations or deletion of the entire domain (Shan and Kruger 1998; Janosik et al. 2001; Evande et al. 2002). The architecture of the regulatory domain and its juxtaposition relative to the catalytic domain was first visualized in the structure of full-length dCBS (Fig. 1.2a). The secondary structures of the two CBS domains are slightly different: $\upbeta \text{--}\upalpha \text{--}\upbeta \text{--}\upbeta \text{--}\upalpha \text{--}\upalpha \text{--}\upbeta$ fold in one (spanning residues 416–468 in hCBS) and α-β-α-β-β-α in the second (residues 486–543). The AdoMet binding sites in CBS is predicted to reside in the β-sheet-lined cleft between the two CBS domains.

Other Mechanisms for Regulation of CBS Activity. As a junction enzyme in sulfur metabolism, CBS is the locus of complex regulation. AdoMet is an allosteric regulator, which activates CBS under conditions of methyl group sufficiency (Prudova et al. 2006; Pey et al. 2013). As noted above, gaseous signaling molecules like NO and CO bind to the heme in CBS and inhibits its activity (Taoka et al. 1999; Taoka and Banerjee 2001; Agarwal and Banerjee 2008; Yamamoto et al. 2011; Yadav et al. 2012), which is reversed upon air oxidation (Kabil et al. 2011b). Sumoylation of CBS decreases its activity (Agarwal and Banerjee 2008). K211 in the catalytic core of human CBS appears to be the site of sumoylation. In addition, the C-terminal regulatory domain of CBS is needed for sumoylation. CBS is reportedly inhibited by lanthionine synthase C-like protein 1 (LanCL1) in the presence of glutathione (Zhong et al. 2012). Under oxidative stress conditions, when oxidized glutathione levels rise, inhibition by LanCL1 is alleviated providing a mechanism for increasing transsulfuration flux and consequently, glutathione synthesis.

1.1.1.2 Cystathionine γ Lyase

Structural Organization of CSE. Human CSE (hCSE) is a homotetrameric enzyme that catalyzes the second step in the transulfuration pathway cleaving cystathionine to cysteine, α-ketobutyrate, and ammonia (Fig. 1.3a). The crystal structures of yeast (yCSE) and hCSE are available at 2.6 Å resolution each (Messerschmidt et al. 2003; Sun et al. 2009). Each hCSE monomer comprises a large N-terminal PLP-binding domain (residues 1–263) and a smaller C-terminal domain (residues 264–401) (Fig. 1.3a). The yCSE monomer comprises three domains: an N-terminal domain, which interacts with the active site of a neighboring monomer, a middle catalytic domain and a small C-terminal domain. The yCSE catalytic and C-terminal domains are organized as in hCSE while the N-terminal extension interacts with the active site of the neighboring subunit, is not present in hCSE and comprises an extended loop, an α-helix and a β-strand (Messerschmidt et al. 2003).

Fig. 1.3

Structure and catalytic mechanism of CSE. (a) Structures of tetrameric hCSE comprising an N-terminal catalytic domain (yellow) and a C-terminal domain (blue). PLP is shown in stick representation (cyan). (b) Close-up of the active site of hCSE. Hydrogen bonds between PLP and amino acids lining the active site are shown as dotted lines. The red labels denote residues contributed by a neighboring subunit. Figure 1.3a, b were generated using PDB file 2NMP. (c) Proposed catalytic cycle of CSE for H₂S generation. Cysteine reacts with the resting enzyme and forms external aldimine of cysteine. Abstraction of the α-proton of bound cysteine leads to formation of cysteine-ketamine intermediate. Cleavage of the C–β–S bond eliminates H₂S and leads to formation of the aminoacrylate intermediate, which undergoes hydrolysis to give pyruvate and ammonia. In the final step, free PLP rebinds with the K212 to regenerate the resting internal aldimine

Mutations in CSE, inherited as an autosomal recessive disorder, results in cystathioninuria, which is often benign (Wang and Hegele 2003). Cystathioninuria can be secondarily associated with hepatoblastoma, neuroblastoma, poor development, cystic fibrosis and Down’s syndrome.

Catalytic Mechanism of CSE. Like CBS, CSE also catalyze various reactions leading to H₂S biogenesis in addition to catalyzing the canonical reaction in the transsulfuration pathway (Chiku et al. 2009; Singh et al. 2009). The PLP is covalently linked to K212 in hCSE via a Schiff base and its mutation to alanine reduces H₂S production ~80-fold compared to wild-type enzyme. Several active site residues are engaged in hydrogen bonding interactions with the PLP (Messerschmidt et al. 2003; Sun et al. 2009); N187 with the pyridine nitrogen and G90, L91, S209 and T211 from one subunit and Y60 and R62 from an adjacent subunit with the phosphate moiety (Fig. 1.3b). Replacement of N187 with alanine or glutamic acid results in complete loss of H₂S production while substitutions of S209 and T211 by alanine result in a modest ~1.5-fold decrease in H₂S production (Huang et al. 2010). Interactions between Y60 and R62 contributed by the N-terminal domain of a neighboring monomer and the phosphate group of PLP helps to stabilize the active site, which is located at the interface between adjacent subunits. Multiple sequence alignment of human, yeast, mouse, rat and slime mold CSEs reveals that Y60 and R62 are conserved in all five sequences (Huang et al. 2010). Substitution of Y60 with threonine or alanine results in an ~5–8-fold decrease in hCSE activity, while replacement of R62 with alanine or lysine results in an ~10–36-fold decrease in activity (Huang et al. 2010).

CSE belongs to the γ-family of PLP-dependent enzymes, which catalyze elimination reactions at the γ-carbon. However, in the case of CSE, the specificity is not high and the enzyme catalyzes reactions at both the β- and γ-carbons of the substrate. In the first step, the substrate (i.e. cystathionine or cysteine) forms a Schiff base with the PLP via a transaldimination reaction, freeing the active site K212 residue. In the next step, K212 presumably acts as the general base and abstracts the α-proton from bound substrate. When cystathionine is the substrate, cleavage of the C-γ-S bond is promoted by a second proton abstraction from Cβ resulting in the subsequent elimination of cysteine. Hydrolysis of the resulting imine intermediate yields α-ketobutyrate and ammonia. Alternatively, when cysteine is the substrate, the C-β-S bond is cleaved, releasing H₂S (Fig. 1.3c). A second transaldimination reaction regenerates the resting internal aldimine. UV-visible spectroscopy based pre-steady state kinetic analysis of the reaction catalyzed by yCSE suggests that product release constitutes the rate-limiting step (Yamagata et al. 2003).

The specificity of the hCSE-catalyzed α,γ-elimination versus α,β-elimination is proposed to be governed by the hydrophobicity of the residue at position 339 (Messerschmidt et al. 2003; Huang et al. 2010). This hypothesis was tested by replacing E339 with lysine, alanine and tyrosine, which increases hydrophobicity in the following order: Y>A>K>E. E339K, E339A and E339Y show approximately 1.8-, 3.2- and 7.2-fold increase, respectively in the catalytic efficiency of H₂S production from cysteine as compared to wild-type enzyme (Huang et al. 2010), consistent with the view that enhancing the hydrophobicity of the residue at position 339 in hCSE favors the α,β-elimination reaction.

Kinetics of H ₂ S Generation by CSE. Detailed steady-state kinetic analysis of hCSE revealed that in addition to the canonical cystathionine cleavage reaction in the transsulfuration pathway (Eq. 1.5), the enzyme can catalyze five distinct H₂S-generating reactions in the presence of cysteine and/or homocysteine (Eqs. 1.6, 1.7, 1.8, 1.9, and 1.10) (Chiku et al. 2009).

$$ \mathrm{ L}\text{--}\mathrm{ Cystathionine} + {{\mathrm{ H}}_2}\mathrm{ O}\to \mathrm{ L}\text{--}\mathrm{ cysteine} + \upalpha \text{--}\mathrm{ Ketobutyrate} + \mathrm{ N}{{\mathrm{ H}}_3} $$

(1.5)

$$ \mathrm{ L}\text{--}\mathrm{ Cysteine} + {{\mathrm{ H}}_2}\mathrm{ O}\to \mathrm{ L}\text{--}\mathrm{ Pyruvate} + \mathrm{ N}{{\mathrm{ H}}_3} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.6)

$$ \mathrm{ L}\text{--}\mathrm{ Cysteine} + \mathrm{ L}\text{--}\mathrm{ Cysteine}\to \mathrm{ L}\text{--}\mathrm{ Lanthionine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.7)

$$ \mathrm{ L}\text{--}\mathrm{ Homocysteine} + {{\mathrm{ H}}_2}\mathrm{ O}\to \mathrm{ L}\text{--}\mathrm{ Homoserine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.8)

$$ \mathrm{ L}\text{--}\mathrm{ Homocysteine} + \mathrm{ L}\text{--}\mathrm{ Homocysteine}\to \mathrm{ L}\text{--}\mathrm{ Homolanthionine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.9)

$$ \mathrm{ L}\text{--}\mathrm{ Homocysteine} + \mathrm{ Cysteine}\to \mathrm{ L}\text{--}\mathrm{ cystathionine} + {{\mathrm{ H}}_2}\mathrm{ S} $$

(1.10)

Unlike hCSE (Chiku et al. 2009), rat CSE reportedly utilizes cystine, the oxidized disulfide form of cysteine, as a substrate for H₂S production (Stipanuk and Beck 1982). However, the availability of cystine in the reducing cellular milieu is questionable and the immediate product of the reaction is cysteine persulfide, which is highly unstable. Under maximal velocity conditions, the highest rate of H₂S production is observed for reaction 1.9, i.e. γ-replacement of homocysteine by a second mole of homocysteine while the lowest rate is observed for reaction 1.6, i.e. α,β−elimination of cysteine (Chiku et al. 2009). At physiological substrate concentrations, the highest rate of H₂S production is predicted to occur via the α,β−elimination reaction of cysteine and the lowest rate from the β−replacement of cysteine by a second mole of cysteine (reaction 1.7).

CSE exhibits two substrate-binding sites: site 1 at which the Schiff base is formed between PLP and an amino acid and site 2 where the nucleophilic second amino acid binds. The K_M for cysteine at site 1 (1.7 ± 0.7 mM) is ~1.6-fold lower than for homocysteine (2.7 ± 0.8 mM) while for the K_M for homocysteine at site 2 (5.9 ± 1.2 mM) is ~6-fold lower than for cysteine (33 ± 8 mM). The lower K_M for cysteine versus homocysteine at site 1 together with the higher cellular concentration of cysteine (~100 μM) versus homocysteine (<10 μM), explains the predominance of reaction 1.6 versus 1.9 at physiologically relevant substrate concentrations (Chiku et al. 2009).

The relative contributions of the various CSE-catalyzed H₂S producing reactions has been estimated at physiologically relevant substrate concentrations (5 μM cystathionine, 100 μM cysteine and 10 μM homocysteine) and low (10 μM), moderate (40 μM) and severe (100 μM) hyperhomocysteinemia (Chiku et al. 2009). Simulations predict that under normal and hyperhomocysteinemic conditions, ~87–99.5 % of the total H₂S is derived via α,β-elimination of cysteine and α,γ-elimination reaction of homocysteine, while the remaining three reactions collectively contribute very little. Between the two major H₂S contributing reactions, the α,β-elimination of cysteine predominates (~71 %) at normal homocysteine concentrations while the α,γ-elimination of homocysteine is the major H₂S producer accounting for ~61 % and ~78 % H₂S, respectively under moderate and severe hyperhomocysteinemic conditions (Chiku et al. 2009).

Preliminary pre-steady state kinetic analyses of the yCSE-catalyzed reaction were performed at 5 °C and 30 °C (Yamagata et al. 2003). An aminocrotonate intermediate was detected (λ_max = 480 nm) upon rapid mixing of yCSE with cystathionine but not with L-cysteine. Conversion of the aminocrotonate intermediate to products represents the rate-limiting step in the CSE-catalyzed cleavage of csytathionine (Yamagata et al. 2003).

Regulation of CSE. Unlike CBS, mechanisms for regulating CSE activity are not well understood. H₂S formation by CSE was reported to be upregulated by calmodulin in the presence of 2 mM Ca²⁺ (Yang et al. 2008). However, Ca²⁺/calmodulin-dependent regulation of purified hCSE has not been observed in our laboratory (Padovani and Banerjee, unpublished results). A recent study on rat CSE supports our observations (Mikami et al. 2013) since the rate of H₂S production was low in the presence of 2 mM cysteine and 0–100 nM Ca²⁺ and increased five-fold when 50 μM PLP was added to the reaction mixture. At higher Ca²⁺ concentrations (0.3–3.0 μM), the rate of H₂S production was ~2.5-fold lower in the presence or absence of PLP compared to that at lower Ca²⁺ concentrations (less than 0.3 μM). Calmodulin (1 μM) had no effect on H₂S production by CSE. These findings suggest that CSE is not regulated by calmodulin and that low Ca²⁺ enhances H₂S production but only in the presence of exogenous PLP while high Ca²⁺ concentrations (300 nM to 3 μM) inhibit CSE (Mikami et al. 2013). While CSE can be sumoylated in vitro (Agarwal and Banerjee 2008), the physiological relevance of this modification is not known. Mammalian CSE has two conserved CXXC motifs. However, their involvement in redox-dependent regulation of CSE activity is not known.

1.1.1.3 Mercaptopyruvate Sulfurtransferase

Structural Organization of MST. MST functions on a catabolic arm of cysteine metabolism and acts downstream of CAT to produce H₂S (Meister et al. 1954; Stipanuk and Beck 1982). CAT, which is identical to aspartate aminotransferase, catalyzes the conversion of cysteine in the presence of α-ketoglutarate to 3-mercaptopyruvate and ammonia. 3-Mercaptopyruvate is a substrate for MST, which transfers the sulfur group to a catalytic cysteine residue forming an enzyme-bound persulfide and releasing pyruvate. In the second half reaction, the sulfane sulfur is transferred to a thiol acceptor e.g. cysteine, homocysteine, dihydrolipoic acid, GSH or thioredoxin and subsequently released as H₂S (Nagahara et al. 2007; Mikami et al. 2011a; Yadav et al. 2013). The MST-bound persulfide can also be transferred to non-thiol acceptors like KCN to form thiocyanate (Nagahara et al. 1999). Recently an alternative to the transamination pathway has been reported for the production of 3-mercaptopyruvate from D-cysteine in a reaction catalyzed by D-amino acid oxidase (Shibuya et al. 2013). The latter, is expressed in multiple tissues and is most abundant in the cerebellum and kidney (Shibuya et al. 2013). Defects in the MST gene are inherited as an autosomal recessive disorder known as mercaptolactate-cysteine disulfiduria (Crawhall et al. 1968). The condition is characterized by excessive excretion of mercaptolactate-cysteine disulfide in the urine, with or without mental retardation. Mercaptopyruvate is oxidized by lactate dehydrogenase to mercaptolactate and reacts subsequent to export, with extracellular cysteine to form the mixed disulfide of mercaptolactate and cysteine (Nagahara and Sawada 2006).

The crystal structures of MST from E. coli and Leishmania major have been solved at 2.8 Å and 2.1 Å resolution, respectively (Alphey et al. 2003; Spallarossa et al. 2004). Recently, our laboratory has obtained the structure of the product complex of hMST at 2.15 Å resolution with pyruvate and persulfide bound at the active site (Yadav et al. 2013). Both bacterial and human MST comprise two domains connected by a linker, which are structurally similar to the rhodanese domain (Fig. 1.4a). The Leishmania MST has an extra ~80 amino acids long domain that shares structural homology with the immunosuppressant FK506-binding protein and to macrophage infectivity potentiator protein, both of which exhibit peptidyl prolyl cis-trans isomerase activity (Alphey et al. 2003). Expression of a C-terminally truncated Leishmania MST yields a misfolded protein devoid of catalytic activity (Williams et al. 2003). The C-terminal domain in the L. major MST is postulated to be involved in protein folding and protein–protein interactions (Alphey et al. 2003).

Fig. 1.4

Structure and catalytic mechanism of MST. (a) Structure of human MST, which comprises an N-terminal domain (yellow) connected to the catalytic domain (blue) by a linker (red). (b) Close-up of the active site of hMST showing a bound persulfide on Cys248 and pyruvate (hydrogen bonds are represented by dotted lines). Figure 1.4a, b were generated using PDB file 4JGT. (c) Reaction scheme for CAT/MST-dependent H₂S generation. (1) CAT catalyzes the transamination between cysteine and α-ketoglutarate to generate mercaptopyruvate and glutamate. (2) MST catalyzes the sulfur transfer from mercaptopyruvate to an active site cysteine, giving pyruvate and MST-bound persulfide. The latter reacts with thiols or thioredoxin (in the presence of NADPH and thioredoxin reductase) to generate H₂S

Catalytic Mechanism of MST. A conserved cysteine residue (C248 in hMST) plays a key role in the catalytic mechanism of this enzyme (Fig. 1.4b). All three available MST structures exhibit persulfides at the active site cysteine residue, which represents the product of the MST-catalyzed sulfur transfer from 3-mercaptopyruvate. Substitution of the corresponding active site cysteine with serine in rat MST (rMST), results in complete loss of activity (Nagahara and Nishino 1996). Two conserved arginine residues (R188 and R197 in hMST) (Fig. 1.4b) are proposed to be important for proper positioning of the substrate and their replacement with glycine is deleterious for rMST activity. The R187G and R196G mutations increase K_M for 3-mercaptopyruvate by 10- and 60-fold, respectively and decrease k _cat 4- and 870-fold, respectively. Based on the Leishmania MST crystal structure, a serine protease-like catalytic triad, comprising Ser255-His75-Asp61 was predicted to be a common feature of the MST family. The catalytic triad is proposed to play a role in polarizing the carbonyl bond in the substrate to assist in the nucleophilic attack by the active site cysteine (Alphey et al. 2003). The crystal structure of hMST confirms the presence of a serine protease-like catalytic triad (S250-H74-D63) in the active site (Yadav et al. 2013).

Enzymes belonging to the sulfurtransferase family catalyze the transfer of a sulfur atom from a sulfur donor to a nucleophilic acceptor. Based on steady-state kinetic analysis of bovine MST, the enzyme was proposed to use a sequential mechanism (Jarabak and Westley 1978). Based on the structure of hMST containing a mixture of the product complex (Cys248-SSH and pyruvate) and an unproductive intermediate (3-mercaptopyruvate in a disulfide linkage with Cys248), we have proposed a detailed reaction mechanism (Fig. 1.4c) (Yadav et al. 2013). The conserved arginine residues, R188 and R197, interact via hydrogen bonds with the carboxyl group of mercaptopyruvate while R197 also hydrogen bonds with the carbonyl group of the substrate. H74 in the catalytic triad is proposed to function as a general base, abstracting a proton from S250, which in turn deprotonates C248 for subsequent attack on the sulfur of 3-mercaptopyruvate, leading to transfer of the sulfur atom to form Cys248-persulfide and pyruvate. Following release of pyruvate, an acceptor molecule (e.g. a dithiol or thioredoxin) attacks the sulfane sulfur of the cysteine persulfide regenerating MST and forming a new persulfide on the acceptor. Nucleophilic attack by the second (or resolving cysteine) on the acceptor releases H₂S and results in oxidized disulfide product. Sulfur transfer from mercaptopyruvate to cyanide generates thiocyanate and pyruvate (Nagahara et al. 1999).

Dihydrolipic acid, thioredoxin, cysteine, homocysteine and glutathione function as acceptors in the in vitro MST assay and a detailed steady state kinetics analysis of H₂S production by MST in the presence of these acceptors have been reported (Yadav et al. 2013). At pH 7.4 and 37 °C kinetic studies revealed that, in the presence of thioredoxin or DHLA, hMST exhibits the highest k _cat/K_M values are obtained in the presence of human thioredoxin (520,000 M⁻¹ s⁻¹) and dihydrolipoic acid (390 M⁻¹ s⁻¹), while the lowest value (12.0 M⁻¹ s⁻¹) was obtained with glutathione. Based on kinetic simulation at physiologically relevant concentrations of acceptors, thioredoxin is predicted to couple most efficiently to the MST reaction (Yadav et al. 2013).

Regulation of MST. Regulation of the MST/CAT pathway is poorly understood. Ca²⁺-dependent regulation of MST has been reported based on the effect of varying Ca²⁺ concentrations on MST/CAT-dependent H₂S production in mouse retinal lysate (Mikami et al. 2011b). H₂S production decreased as Ca²⁺ concentration increased (0–2.9 μM) in the presence of cysteine and α-ketoglutarate (substrates for the CAT/MST pathway) but not when 3-mercaptopyruvate was used, suggesting that Ca²⁺ regulates CAT (Mikami et al. 2011b). Calmodulin is not involved in regulation of MST/CAT-dependent H₂S production (Mikami et al. 2011b). Rat MST has five cysteines and appears to be redox regulated. Of the five cysteines, three (i.e. C154, C247 & C263) are surface exposed (Nagahara and Katayama 2005; Nagahara 2012). An intersubunit disulfide bond forms between C154 and C263 under oxidizing conditions and can be reduced by thioredoxin. Thioredoxin-reduced MST is ~4.6-fold more active than the oxidized form of MST, while pretreatment of MST with DTT results in lower activation (~2.3-fold) (Nagahara et al. 2007). Human MST is a monomer and the cysteine residues that form the intersubunit disulfide in rMST are not conserved in hMST (Yadav et al. 2013). The active site cysteine (i.e. C247 in rMST) acts as another redox-sensitive switch with the potential to regulate MST (Nagahara and Katayama 2005), Treatment of rMST with stoichiometric oxidants (H₂O₂ or tetrathionate) results in inhibition due to formation of cysteine sulfenate at the active site, which can be reversed by reductants such as DTT or thioredoxin (Nagahara and Katayama 2005).

1.1.1.4 The Relative Contributions of CBS, CSE and MST to H₂S Production

It is not readily possible from the available kinetic data collected under varied buffer, pH and temperature conditions, to assess the relative roles of CBS, CSE and MST to H₂S production in different tissues. As a first step towards addressing this question, our laboratory has initiated kinetic studies in a limited set of tissues (murine liver, kidney and brain) at pH 7.4 and at 37 °C (Kabil et al. 2011; Vitvitsky et al. 2012). A second key piece of information that is needed to evaluate the contributions of the individual enzymes, is their concentrations in a given tissue, which can be obtained using quantitative Western blot analysis (Kabil et al. 2011). Finally, it is essential that sensitive and reliable methods for H₂S detection be used to monitor its formation at physiologically relevant substrate concentrations. Initial studies in our laboratory evaluating total H₂S production in murine tissues indicates that in the presence of 100 μM cysteine, liver exhibits the highest rate of H₂S production (484 ± 271 μmole h⁻¹ kg⁻¹ tissue) followed by kidney (104 ± 44 μmole h⁻¹ kg⁻¹ tissue at 0.5 mM cysteine) and then brain (29 ± 7 μmole h⁻¹ kg⁻¹ tissue) (Kabil et al. 2011). Quantitative Western blot analyses suggest that the expression level of all three H₂S-producing enzymes decreases in the following order: liver>kidney>brain (Kabil et al. 2011; Yadav and Banerjee, unpublished results).

The relative contributions of CBS and CSE to H₂S production at physiologically relevant substrates concentrations (560 μM serine, 100 μM cysteine and varying homocysteine ranging from its normal concentration (10 μM), to those seen under mild (40 μM) and severe (200 μM) hyperhomocysteinemia conditions) have been assessed. The kinetic simulations predict that CSE is the major H₂S producer and accounts for ~97 % of H₂S production in liver at 10 μM homocysteine, with the proportion only increasing under hyperhomocysteinemic conditions (Kabil et al. 2011). A comparable analysis including the contribution of MST is needed to provide a more complete picture of the quantitative significance of CBS, CSE and MST to H₂S production in different tissues. To assess the contribution of the MST/CAT pathway to cysteine-derived H₂S production, the assays need to be conducted in the presence of physiological concentrations of α-ketoglutarate needed by CAT to convert cysteine to 3-mercaptopyruvate.

1.1.2 Enzymology of H₂S Oxidation

1.1.2.1 Microbial Strategies for Sulfide Oxidation

A major reaction of the global sulfur cycle is the oxidation of hydrogen sulfide to sulfate. It is instructive to examine the variations in sulfide oxidation strategies used by microbes where the pathways are much better understood than in man. In microbes, inorganic sulfur compounds such as sulfide, sulfur globules, sulfite, thiosulfate and polythionates are oxidized to sulfate for generation of ATP.

Sulfide and Thiosulfate Oxidation by Phototrophic Bacteria. Green and purple sulfur bacteria utilize both sulfide and thiosulfate as electron donors for photoautotrophic growth (Gregersen et al. 2011; Grimm et al. 2011). Oxidation occurs in a stepwise manner with sulfide being converted initially to elemental sulfur, which is stored as sulfur globules either in the periplasm or on the surface of the outer membrane and utilized when sulfide is limiting. The two major enzymes directly involved in sulfide oxidation are SQR (Eq. 1.11) and flavocytochrome c sulfide dehydrogenase (Eq. 1.12), a heterodimeric flavoprotein comprising a glutathione reductase-like subunit containing an FAD cofactor and a redox-active disulfide and a diheme cyctochrome c subunit (Chen et al. 1994). Flavocytochrome c sulfide dehydrogenase exists in both soluble and membrane bound forms and, like SQR, catalyzes the oxidation of sulfide to polysulfide while reducing the diheme cytochrome via FAD (Chen et al. 1994).

$$ {{\mathrm{ S}}^{2- }} + \mathrm{ Ubiquinon}{{\mathrm{ e}}_{\mathrm{ ox}}}\to {{\mathrm{ S}}^0} + \mathrm{ Ubiquinon}{{\mathrm{ e}}_{\mathrm{ red}}} $$

(1.11)

$$ {{\mathrm{ S}}^{2- }} + \mathrm{ cyt}{{\mathrm{ c}}_{\mathrm{ ox}}}\to\ {{\mathrm{ S}}^0} + \mathrm{ cyt}{{\mathrm{ c}}_{\mathrm{ red}}} $$

(1.12)

Sulfur globules are converted to sulfite via the cytoplasmic dissimilatory sulfite reductase pathway (Holkenbrink et al. 2011). The proteins common to bacteria harboring the dissimilatory sulfite reductase pathway include a siroheme or siroamide-containing sulfite reductase, a transmembrane electron transporting protein complex, a putative siroheme amidase and the product of the dsrC gene. In addition, a sulfurtransferase complex and an iron-sulfur cluster-containing NADH oxidoreductase, which is proposed to reduce sulfur polysulfides, are also present in some bacteria. The transmembrane complex shuttles electrons released during oxidation of H₂S to sulfite to quinones (Gregersen et al. 2011; Holkenbrink et al. 2011). Sulfite is toxic and is further oxidized to sulfate by one of two pathways. The first is direct oxidation catalyzed by sulfite oxidase (Eq. 1.13) and the second involves oxidation of sulfite to adenosine-5′-phosphosulfate and its subsequent conversion to sulfate liberating either ADP or ATP as the co-product (Eqs. 1.14 and 1.15) (Kappler and Dahl 2001).

$$ \mathrm{ S}{{\mathrm{ O}}_3}^{2-} + {{\mathrm{ H}}_2}\mathrm{ O}\to \mathrm{ S}{{\mathrm{ O}}_4}^{2-} + 2{{\mathrm{ e}}^{-}} + 2{{\mathrm{ H}}^{+}} $$

(1.13)

$$ \mathrm{ S}{{\mathrm{ O}}_3}^{2-} + \mathrm{ AMP}\to \mathrm{ APS} + 2{{\mathrm{ e}}^{-}} $$

(1.14)

$$ \mathrm{ APS} + \mathrm{ PPi}(\mathrm{ or}\ \mathrm{ Pi})\to \mathrm{ S}{{\mathrm{ O}}_4}^{2-} + \mathrm{ ATP}(\mathrm{ or}\ \mathrm{ ADP}) $$

(1.15)

Thiosulfate is oxidized to sulfate by the multi-enzyme sulfur oxidizing (Sox) system (Eq. 1.16). The Sox system is widely distributed among phototrophic and sulfur oxidizing bacteria and also present in some green and purple bacteria

$$ ^{-}\mathrm{ SS}{{\mathrm{ O}}_3} + 5{{\mathrm{ H}}_2}\mathrm{ O}\to 2\mathrm{ S}{{\mathrm{ O}}_4}^{2-} + 8{{\mathrm{ e}}^{-}} + 10{{\mathrm{ H}}^{+}} $$

(1.16)

(Friedrich et al. 2001; Gregersen et al. 2011; Grimm et al. 2011). It is localized in the periplasm and comprises four proteins: SoxAX, SoxYZ, SoxB and SoxCD. In the first step, thiosulfate is oxidized to a cysteinyl S-thiosulfonate intermediate bound to SoxY, which is then hydrolyzed to sulfate by a dimanganese thiosulfohydrolase, SoxB. The remaining protein-bound sulfane sulfur is oxidized to sulfate by SoxCD, a heterotetramer containing a molybdenum cofactor-containing subunit and a diheme cytochrome c-containing subunit. Electrons released during thiosulfate oxidation are transferred to cytochromes and from there, to the electron transport chain. Green and purple sulfur bacteria do not have a SoxCD component in their Sox system, and are unable to completely oxidize thiosulfate using the Sox system (Sakurai et al. 2010; Gregersen et al. 2011; Grimm et al. 2011). In these bacteria, the remaining sulfane sulfur in the SoxYZ complex is proposed to either be transferred to the bacterial sulfur globule pool or, to be oxidized by the dissimilatory sulfite reductase pathway to sulfite.

Sulfur Oxidation by Acidophilic Bacteria . Acidophilic bacteria such as Acidithiobacillus ferrooxidans and Acidithiobacillus thiooxidans, utilize a sulfur dioxygenase to oxidize elemental sulfur to sulfite (Eq. 1.17) (Rohwerder and Sand 2003). Thiol-containing proteins are predicted to be involved in mobilization of extracellular octameric elemental sulfur and presentation of sulfane sulfur to the

$$ ^1{/_8}{{\mathrm{ S}}_8} + {{\mathrm{ O}}_2} + {{\mathrm{ H}}_2}\mathrm{ O}\to \mathrm{ S}{{\mathrm{ O}}_3}^{2\rm{-}} + 2{{\mathrm{ H}}^{+}} $$

(1.17)

periplasmic sulfur dioxygenase, which does not accept sulfide or elemental sulfur as substrate (Rohwerder and Sand 2003). The product of this reaction, sulfite is oxidized to sulfate by sulfite oxidase. Two other enzymes, SQR and thiosulfate quinone oxidoreductase are also active in sulfur oxidation in acidophilic bacteria. SQR oxidizes sulfide to elemental sulfur which can react abiotically with sulfite to produce thiosulfate (Rohwerder and Sand 2007). Thiosulfate can be oxidized to tetrathionate by thiosulfate quinone oxidoreductase with the electrons being transferred to quinones.

Sulfur Oxidation by Archaea. Sulfur-oxidizing Achaea use a variation of the oxidation schemes discussed above. Thus, in Acidianus ambivalens, sulfur oxidation is initiated by sulfur oxygenase reductase a non-heme iron protein that catalyzes the disproportionation of elemental sulfur and/or polysulfide to sulfide and sulfite (Kletzin 1992; Kletzin et al. 2004; Urich et al. 2006). H₂S is oxidized by SQR to elemental sulfur, which can be utilized by sulfur oxygenase reductase setting up an energy-yielding cycle between the two enzymes and allowing the organism to maximize the energy gained from sulfur compounds (Brito et al. 2009). Sulfite can be further oxidized to sulfate by sulfite oxidase (Kletzin et al. 2004; Brito et al. 2009) or it can react with elemental sulfur to form thiosulfate, which is oxidized to tetrathionate. Alternatively, sulfite can be oxidized to sulfate via the APS pathway (Zimmermann et al. 1999).

In the next section, we review the literature on the mitochondrial sulfide oxidation pathway focusing on the enzymes, SQR, the persulfide dioxygenase (ETHE1) and rhodanese. The conversion of sulfide to sulfite and thiosulfate by this trio of enzymes most closely resembles the components of the sulfide oxidation pathway in acidophilic bacteria with the exception that polysulfide is not generated as an intermediate in the mammalian pathway. Instead, as discussed below, the persulfide product of SQR is transferred to an as yet unidentified acceptor. Sulfite is ultimately oxidized to sulfate by sulfite oxidase in mammals.

1.1.2.2 Sulfide Quinone Oxidoreductase

Structural Organization of SQR. SQR is found in the inner mitochondrial membrane (Theissen et al. 2003) in eukaryotes, which are believed to have acquired the nuclear encode gene from a mitochondrial endosymbiont (Theissen et al. 2003). SQR exists as a dimer or a trimer, with one FAD and one redox active disulfide in each monomer (Marcia et al. 2010). SQR oxidizes sulfide to a protein-bound persulfide and transfers electrons from H₂S to ubiquinone via a bound FAD, coupling sulfide oxidation to the electron transport chain (Fig. 1.1). Hence, sulfide functions as an inorganic substrate for the ATP-generating electron transfer chain (Bouillaud and Blachier 2011).

Several crystal structures of SQRs have been reported, notably from Acidianus ambivalens (Brito et al. 2009), Aquifex aeolicus (Marcia et al. 2009), and Acidithiobacillus ferrooxidans (Cherney et al. 2010). SQR belongs to the family of flavin disulfide reductases that includes glutathione reductase. SQR contains two Rossman fold domains and a C-terminal domain that is important for membrane binding (Fig. 1.5a). The FAD cofactor is found in the first N-terminal Rossman fold domain and can be non-covalently or covalently bound. In the latter case, a thioether linkage exists between a cysteine or a cysteine persulfide and the 8-methylene group of the isoalloxazine ring of FAD (Marcia et al. 2009). The catalytic disulfide is located on the re face of the FAD. A conserved glutamate residue in the active site is proposed to serve as a general base for deprotonating H₂S (Cherney et al. 2010). The ubiquinone-binding site is located on the si face of the FAD and the majority of residues in contact with the quinone are hydrophobic including F41, P43, G322, Y323, N353, Y411, F394 and F357 (A. ferrooxidans numbering) (Cherney et al. 2010). The aromatic ring of the quinone is sandwiched between the benzene rings of F394 and F357. The O2 atom of the flavin electron donor, and the O4 atom of the quinone acceptor are <3 Å apart. Residues Y411 and K391 have been proposed to transfer protons from water for protonation of the reduced quinone (Marcia et al. 2009; Cherney et al. 2010).

Fig. 1.5

Structure and reaction mechanism of SQR. (a) Structure of A. ferrooxidans SQR monomer displaying the N-terminal Rossman fold domains (blue) and the C-terminal membrane binding domain (yellow) and the cofactors, FAD (yellow) and DUQ (green). Active site sulfurs are shown as gold spheres (b) Close-up of the A. ferrooxidans SQR active site viewing the re face of the FAD isoalloxazine ring and decylubiquinone (DUQ) on the si face of FAD. Sulfur atoms between the active site cysteine residues are shown as gold spheres. Figure 1.5a, b were generated using PDB file 3T31. (c) Proposed mechanism of SQR. Reaction of H₂S reduces with the active site disulfide bond results in the formation of a persulfide on one cysteine and a covalent linkage between the other cysteine and FAD. The cysteine bound persulfide is released from the enzyme by an acceptor molecule. Reformation of the active site disulfide results in reduction of FAD to FADH₂. Electrons are then transferred from FADH₂ to ubiquinone allowing for regeneration of the oxidized active site

Catalytic Mechanism of SQR. The catalytic cycle of SQR is initiated by nucleophilic attack of the sulfide on the disulfide resulting in formation of a persulfide and a cysteine thiolate (Fig. 1.5b) (Brito et al. 2009; Marcia et al. 2009; Cherney et al. 2010, 2012; Jackson et al. 2012). The cysteine thiolate attacks the FAD cofactor forming a C4A adduct (Fig. 1.5c). Nucleophilic attack of an acceptor on the sulfane sulfur results in reformation of the active site disulfide and subsequent two-electron transfer to the FAD results in formation of FADH₂. Electron transfer from FADH₂ to ubiquinone regenerates FAD.

In bacterial SQRs, multiple rounds of electron transfer occur from sulfide to quinone without release of the persulfide at the end of each catalytic cycle (Brito et al. 2009; Marcia et al. 2009; Cherney et al. 2010; Jackson et al. 2012). Instead, the catalytic cycle repeats until the maximum length of the polysulfide product that can be accommodated in the active site is obtained at which point two consecutive nucleophilic attacks by the sulfide results in product release and reformation of the active site disulfide. Snapshots of these polysulfide species have been observed in structures of prokaryotic SQRs. The structure of A. ambivalens SQR revealed a trisulfide bridge between the two active site cysteines (Brito et al. 2009), while the A. ferrooxidans structure revealed a branched intermediate between the two cysteine residues consisting of five sulfur atoms (Cherney et al. 2010, 2012). In the A. aeolicus SQR structures, both linear and cyclic polysulfur intermediates were observed (Marcia et al. 2009). The released product can be soluble polysulfide containing up to ten sulfur atoms (Griesbeck et al. 2002) or an octasulfur ring (Marcia et al. 2009).

In contrast, in mammalian SQRs, the persulfide product is transferred to an acceptor at the end of each catalytic cycle. Although several molecules have been proposed as acceptors of the persulfide product including sulfite and glutathione, the identity of the physiological co-substrate is not known (Jackson et al. 2012). Human SQR utilizes cyanide and sulfite as persulfide acceptors and exhibits k _cat values of 82 ± 6 s⁻¹ and 251 ± 9 s⁻¹ in the presence of 1 mM cyanide and 600 μM sulfite respectively (Jackson et al. 2012). In contrast, the k _cat in the presence of 1 mM glutathione as an acceptor is 19 ± 3 s⁻¹, which is reportedly identical to the background rate constant observed in the absence of an additional acceptor (k _cat = 18.5 ± 0.9 s⁻¹). The activity of SQR in the absence of an exogenous acceptor was attributed to sulfide acting as an acceptor for the SQR product resulting in formation of hydrogen disulfide (Jackson et al. 2012). It is not known whether ETHE1 can directly accept the persulfide product from SQR.

1.1.2.3 Persulfide Dioxygenase (ETHE1)

ETHE1, a persulfide dioxygenase, is a soluble, mitochondrial matrix enzyme that catalyzes the oxidation of glutathione persulfide to sulfite (Hildebrandt and Grieshaber 2008; Tiranti et al. 2009; Kabil and Banerjee 2012). However, the physiological substrate is not known unequivocally. Persulfide dioxygenase contains a mononuclear non-heme iron in its active site (McCoy et al. 2006; Holdorf et al. 2008). Mutations in persulfide dioxygenase result in ethylmalonic encephalopathy, an autosomal recessive disorder that results in developmental delay hemorrhagic diarrhea, acrocyanosis, petechiae, and progressive neurological failure (Tiranti et al. 2004, 2009; Mineri et al. 2008). The clinical profile of ethylmalonic encephalopathy includes high levels of lactate, high C5 and C4 acylcarnitine levels in blood, increased ethylmalonic acid concentrations in urine and cytochrome c oxidase deficiency in muscle and brain (Burlina et al. 1991; Garcia-Silva et al. 1994; Tiranti et al. 2009). Persulfide dioxygenase deficiency leads to accumulation of thiosulfate and sulfide with the latter inhibiting cytochrome c oxidase and possibly other enzymes and accounting in part for the observed pathology (Di Meo et al. 2011).

Both glutathione persulfide (GSSH) and CoA persulfide (CoA-SSH) serve as substrates for human persulfide dioxygenase while cysteine persulfide, glutathione and thiosulfate do not (Kabil and Banerjee 2012). However, the specific activity of the enzyme with CoA-SSH is ~50-fold lower than with GSSH. Interestingly, 5 mM glutathione, a high but physiologically relevant concentration, increases the catalytic efficiency of the persulfide dioxygenase ~3-fold. Despite the similarity between persulfide dioxygenase and glyoxylase II, the former is unable to hydrolyze S-D-lactoylglutathione and other glutathione thioesters like glyoxylase II (Holdorf et al. 2008). This is most likely due to a C-terminal loop covering much of the active site in the dioxygenase making it smaller than the glyoxylase II active site (McCoy et al. 2006; Holdorf et al. 2008).

Structural Organization of Persulfide Dioxygenase . The structure of the Arabidopsis thaliana persulfide dioxygenase (McCoy et al. 2006) has been used to model the structure of the human enzyme (Fig. 1.6a) (Kabil and Banerjee 2012). The two proteins share 54 % identity. However, while the Arabidopsis persulfide dioxygenase is a dimer, as revealed by gel-filtration chromatography and crystallography (McCoy et al. 2006), the human enzyme is a monomer (Kabil and Banerjee 2012). The Arabidopsis persulfide dioxygenase has a typical metallo-β-lactamase fold containing two central β-sheets enclosed by three helices on each side. The active site iron is ligated via two histidine and one aspartate residue, which correspond to H84, H135 and D154 in the human sequence (Fig. 1.6b). The three remaining coordination sites are occupied by water resulting in octahedral coordination, typical for ferrous iron (McCoy et al. 2006; Holdorf et al. 2008).

Fig. 1.6

Structure and catalytic mechanism of persulfide dioxygenase. (a) Structure of A. thaliana persulfide dioxygenase monomer. The mononuclear non-heme iron is shown in orange and active site residues are shown in cyan. (b) Close up of the active site of A. thaliana persulfide dioxygenase. The mononuclear non-heme iron is coordinated by the 2His-1Asp facial triad residues (blue) and other active site residues (cyan). Figure 1.6a, b were generated using PDB file 2GCU. (c) Proposed catalytic mechanism of persulfide dioxygenase for sulfite generation. Binding of the GSSH substrate displaces coordinated water and promotes binding of oxygen to the iron center to form the Fe (III) superoxo intermediate. Resonance allows for partial radical cation character of the coordinated sulfur leading to recombination and formation of a cyclic peroxo-intermediate. Cleavage of the O-O bond results in a sulfoxy-cation and an iron-bound activated oxygen atom which is transferred to the sulfur to sulfoxy-cation. Subsequent hydrolysis yields sulfite, and GSH is displaced from the active site upon water binding to the metal center. R ₁ and R ₂ represent residues glutamate and glycine in the GSSH substrate

The geometry of the iron site in persulfide dioxygenase resembles that of the 2-His-1-carboxylate facial triad family of oxygenases (Hegg and Que 1997), a common structural motif that binds mononuclear non-heme Fe²⁺. The three coordination sites occupied by water are available for binding ligands such as O₂, substrates and/or cofactors, and allow the enzymes to tune the reactivity of the iron center (Koehntop et al. 2005; Bruijnincx et al. 2008). With three coordination sites occupied by solvent, the metal center is not reactive towards dioxygen. Substrate binding displaces solvent molecules and results in formation of a five-coordinate metal center. Replacement of neutral solvent molecules with an anionic ligand can decrease the Fe(III)/Fe(II) redox potential rendering the iron center more susceptible to oxidation by dioxygen (Koehntop et al. 2005). Hence, substrate binding primes the iron center for dioxygen binding and protects the enzyme from auto-inactivation (Koehntop et al. 2005).

The substrate-binding site is predicted to comprise residues M226, R163, Y197 and F166 in the human sequence based on structural alignment with the active site of glyoxylase II, which provides insights into how the glutathione persulfide might interacts with persulfide dioxygenase. The glyoxylase II residues Y145 and K143 are engaged in hydrogen bonds with the glutamate portion of glutathione and Y175 forms a hydrogen bond with the cysteine portion of glutathione (McCoy et al. 2006). The corresponding residues in human persulfide dioxygenase are F166, R163 and Y197. R163 is predicted to be positioned similarly to K143 and to interact with the glutamate portion of the GSSH substrate. In contrast, F166, which replaces Y145 in the persulfide dioxygenase structure, is not expected to engage in a hydrogen bonding interaction with the substrate. Y197 like Y175 in the A. thaliana protein, might interact with the cysteine portion of GSSH, while the sulfane sulfur of GSSH would be coordinated to the iron center (Kabil and Banerjee 2012).

Catalytic Mechanism of Persulfide Dioxygenase. Kinetic characterization of human persulfide dioxygenase and two variants mimicking missense mutations described in ethylmalonic aciduria patients, T151I and D196N, has been described (Kabil and Banerjee 2012). In the presence of GSSH, the V_max for human persulfide dioxygenase is 113 ± 4 μmol min⁻¹ mg⁻¹ protein, which corresponds to a k _cat of 47 s⁻¹at 22 °C. Glutathione (5 mM) decreases the K_M for GSSH ~1.4 fold and increases k _cat an ~2.2-fold yielding an ~3-fold increase in k _cat/K_M. The physiological relevance of modulation of the persulfide dioxygenase activity by GSH is not known. The T152I mutant does not affect the K_M for GSSH while the k _cat is diminished ~4-fold and correlates with an ~2.5 fold lower iron content compared to wild-type enzyme. T152 is located on the same loop as D154, a predicted iron ligand and participates in a hydrogen bonding interaction with the backbone of L156, which may be necessary for correct positioning of D154. The T152I mutation would result in loss of the interaction with L156 possibly leading to repositioning of D154 (Kabil and Banerjee 2012).

The impact of the D196N mutation is to increase the K_M for GSSH ~2-fold while leaving the k _cat unaltered. D196 is located on an internal loop distal from the active site and is proposed to hydrogen bond with F200 and H198 to stabilize the loop. The D196N mutation likely destabilizes this loop. Several other pathogenic missense mutations have been described in human persulfide dioxygenase and include Y38C, L55P, T136A, R163Q, R163W, C161Y, T164K and L185R (Tiranti et al. 2004, 2006; Mineri et al. 2008). Many of these mutations are predicted to be located near the active site and may disrupt metal or substrate binding and/or destabilize the enzyme.

The reaction mechanism proposed for persulfide dioxygenase is adapted from that described for cysteine dioxygenase (McCoy et al. 2006) and is also based on the general reaction mechanism of mononuclear non-heme iron oxygenases (Koehntop et al. 2005; Bruijnincx et al. 2008). In the proposed mechanism (Kabil and Banerjee 2012), solvent is displaced from the iron upon GSSH binding (Fig. 1.6c). The sulfane sulfur and a nitrogen atom from GSSH coordinate to the iron center resulting in a five-coordinate iron species, which is primed for O₂ binding. Binding of O₂ results in formation of a Fe (III) superoxo intermediate in which the coordinated sulfur acquires a partial radical cation character via resonance. Recombination of the Fe(III) superoxo species with the coordinated sulfur atom leads to formation of a cyclic peroxo-intermediate and is followed by homolytic cleavage of the O-O bond resulting in a sulfoxy-cation species and a metal-bound activated oxygen atom. Transfer of the activated oxygen to the sulfur and hydrolysis yields sulfite. Finally, release of GSH followed by rebinding of solvent to the active site completes the catalytic cycle (Kabil and Banerjee 2012).

1.1.2.4 Rhodanese

Rhodanese is a widely distributed protein found in Archaea, bacteria and eukaryotes (Cipollone et al. 2007). It is a mitochondrial matrix protein that catalyzes the transfer of a sulfur atom from a sulfur donor to an acceptor (Eqs. 1.18 and 1.19). Despite extensive studies on rhodanese, its precise physiological function is still not known. Historically, rhodanese was thought to be involved in

$$ \mathrm{ S}{{\mathrm{ O}}_3}^{2-} + \mathrm{ RSSH}\to {{\mathrm{ S}}_2}{{\mathrm{ O}}_3}^{2-} + \mathrm{ RSH} $$

(1.18)

$$ {{\mathrm{ S}}_2}{{\mathrm{ O}}_3}^{2-} + \mathrm{ C}{{\mathrm{ N}}^{-}}\to \mathrm{ S}{{\mathrm{ O}}_3}^{2-} + \mathrm{ S}\mathrm{ C}{{\mathrm{ N}}^{-}} $$

(1.19)

cyanide detoxification due to its ability to convert cyanide and thiosulfate to thiocyanate (Westley 1973; Cipollone et al. 2007). In fact, thiosulfate along with sodium nitrite is administered to treat acute cyanide poisoning. Cells that are routinely exposed to gaseous and dietary intake of cyanide, such as the epithelial cells surrounding bronchioles, hepatocytes that are proximal to the liver’s blood supply and proximal tubule cells in kidney, have the highest rhodanese levels (Sylvester and Sander 1990). However, the activity of rhodanese is confined to the mitochondrial matrix, where thiosulfate enters with low efficiency suggesting that a different sulfur source than thiosulfate might be used for clearing cyanide (Westley et al. 1983). Alternatively, rhodanese might play a role in sulfur metabolism (Westley et al. 1983; Hildebrandt and Grieshaber 2008), particularly for mitochondrial thiosulfate production by transfer of the sulfane sulfur from a donor to sulfite (Hildebrandt and Grieshaber 2008).

Structural Organization of Rhodanese. Bovine liver rhodanese, which shares 89 % sequence identity with the human enzyme, has been characterized extensively. This protein is folded into two globular domains of equal size, with each containing a five stranded parallel β-sheet enclosed by two α-helices on one side and three α-helices on the other (Fig. 1.7a) (Hol et al. 1983). While both domains have a similar fold, the sequence homology is low with only 16 % sequence identity between them. The N-terminal domain is inactive due to replacement of the catalytic cysteine residue with an aspartate and plays a role in forming the active site. The catalytic C247 residue in bovine liver rhodanese is located at the bottom of a pocket formed between the two domains. The walls of the pocket comprise hydrophobic and hydrophilic regions (Fig. 1.7b). The hydrophobic regions consists of residues F212, F106, Y107, W35, and V251, and the hydrophilic region comprise residues D180, S181, R182, R186, E193, R248, K249, and T252. The persulfide intermediate is stabilized by hydrogen bonds from the backbone amides of R248, K249, V251 and the hydroxyl group of T252. Residues R186 and K249 located at the entrance of the active site pocket and their side chains are positioned to participate in binding and positioning of thiosulfate through ionic interactions and to polarize the sulfane sulfur for nucleophilic attack by the C247 thiol. It is possible that the hydrophobic region lining the active site is important for binding other substrates that contain aromatic or hydrophobic residues (Hol et al. 1983).

Fig. 1.7

Structure and reaction mechanism of rhodanese. (a) Structure of Bos taurus rhodanese. The protein comprises two equal sized N-terminal (yellow) and C-terminal (blue) globular domains. The linker region between the two domains is shown in cyan. (b) Close up of the rhodanese active site. The active site is located at the interface of the two domains. N- and C-terminal domain residues contribute to a hydrophobic patch (yellow and blue), while C-terminal domain residues contribute to a hydrophilic region (blue). Figure 1.7a, b were generated using PDB file 1RHD. (c) Catalytic mechanism of rhodanese for thiocyanate generation from thiosulfate. Rhodanese catalyzes the sulfur transfer from thiosulfate to an active site cysteine, resulting in sulfite and a rhodanese-bound persulfide intermediate. The latter reacts with a sulfur acceptor, in this example cyanide, to generate thiocyanate

Catalytic Mechanism of Rhodanese. The sulfurtransferase reaction catalyzed by rhodanese occurs via a double displacement mechanism and involves formation of a stable persulfide intermediate (Hol et al. 1983; Cipollone et al. 2007a). The reaction is initiated by the nucleophilic attack of the C247 thiolate on the sulfane sulfur of thiosulfate resulting in formation of an enzyme-bound persulfide intermediate (Fig. 1.7c). Next, sulfite is released and followed by binding of a sulfur acceptor, which attacks the sulfane sulfur of the persulfide intermediate. In the case of cyanide, the product of this sulfur transfer reaction is thiocyanate (Hol et al. 1983). Release of thiocyanate completes the catalytic cycle.

Rhodanese also displays sulfurtransferase activity between a persulfide donor and sulfite acceptor (Eq. 1.18). The K_M values for cyanide and thiosulfate for bovine rhodanese are 0.087 ± 0.009 mM and 16.2 ± 1.6 mM, respectively (Hildebrandt and Grieshaber 2008). Similar values have also been reported for rat and lugworm rhodanese (Hildebrandt and Grieshaber 2008). In contrast, the K_Ms for GSSH and sulfite are considerably lower (61.3 ± 17.8 μM and 21.8 ± 3.6 μM respectively), suggesting that the persulfide transferase activity of rhodanese might be more relevant physiologically than cyanide detoxification.

1.1.2.5 Mitochondrial Sulfide Oxidation: Unanswered Questions

Endogenously produced H₂S must be regulated to maintain low intracellular concentrations. In mammals, steady-state levels of H₂S are governed by flux through the synthetic pathways (transsulfuration and CAT/MST) and the sulfide oxidation pathway (Vitvitsky et al. 2012). In some prokaryotes, sulfide oxidation is essential for ATP generation. Similarly, low concentrations of H₂S (0.1–1 μM), can stimulate mammalian mitochondrial ATP production and serve as an inorganic source of ATP (Goubern et al. 2007; Bouillaud and Blachier 2011; Modis et al. 2013). An important unanswered question regarding the sulfide oxidation pathway is its organization. Since SQR oxidizes sulfide, it is reasonable to propose that it catalyzes the first step in the pathway. The ambiguity arises thereafter since the persulfide acceptor of SQR is not known. Both persulfide dioxygenase and rhodanese can oxidize persulfide forming either sulfite or thiosulfate, respectively. However, the co-substrate for rhodanese is sulfite, which is derived from the persulfide dioxygenase-catalyzed reaction. This dependence of rhodanese on the product of the dioxygenase suggests that oxidation of H₂S proceeds through SQR, the dioxygenase and then rhodanese. The products of this pathway thus configured are thiosulfate and sulfate, which is derived by oxidation of sulfite catalyzed by sulfite oxidase. The presence of two routes for sulfite oxidation in the mitochondria is paralleled in microbes where multiple sulfite oxidation routes co-exist in the same organism, e.g. the oxidation of sulfite by sulfite oxidase or via the adenosine-5′-phosphosulfate reductase pathway (Kappler and Dahl 2001).

Variants of persulfide dioxygenase fused to a rhodanese domain are found in certain bacteria (Tiranti et al. 2009). The occurrence of fused persulfide dioxygenase/rhodanese variants suggests that their proximity enhances utilization of sulfite produced by one active site and consumed by the other. However, this order of the pathway is brought into question by clinical data on ETHE1 deficient patients and ETHE1 knockout mice in which sulfite levels are greatly diminished, as expected, but thiosulfate and H₂S levels are elevated (Tiranti et al. 2009). Elevated thiosulfate in the absence of persulfide dioxygenase activity suggests that an alternative route for sulfite synthesis exists, which supports production of thiosulfate by rhodanese (Kabil and Banerjee 2012). One branch of the cysteine catabolic pathway is initiated by cysteine dioxygenase that oxidizes cysteine to cysteinesulfinate, which is further metabolized to 3-sulfinylpyruvate by CAT. 3-sulfinylpyruvate is unstable and decomposes to form pyruvate and sulfite. We have posited that the persulfide product of SQR is preferentially utilized by rhodanese under conditions of persulfide dioxygenase deficiency, explaining the observed accumulation of thiosulfate under these conditions (Kabil and Banerjee 2012). However, while cysteine catabolism is up regulated under conditions of cysteine excess (Stipanuk et al. 2004), it is unclear how this pathway responds to persulfide dioxygenase deficiency and its role in sulfite production under these conditions warrants investigation.

In this context, understanding the fate of the persulfide product of SQR is pertinent. Bacterial polysulfide products of SQR are stored in sulfur globules until further oxidation to sulfate. However, sulfur globules have not been reported in higher organisms and in fact, eukaryotic lugworm, rat, and human SQRs require a persulfide acceptor for catalytic turnover under in vitro conditions (Hildebrandt and Grieshaber 2008; Jackson et al. 2012). The persulfide product bound to the SQR active site might be either oxidized directly by persulfide dioxygenase or rhodanese or transferred to a small molecule carrier such as GSH with the resulting GSSH serving as substrate for the persulfide dioxygenase and rhodanese. Sulfane sulfur acceptors that support the activity of SQR include sulfite, cyanide, sulfide and glutathione. Human SQR has been proposed to utilize sulfite as the physiological acceptor since it displays a 4- or 13-fold higher catalytic efficiency in the presence of either cyanide or sulfite versus glutathione (Jackson et al. 2012). Based on these results, it has been proposed that sulfite is the physiological acceptor of SQR’s persulfide product resulting in the formation of thiosulfate (Jackson et al. 2012). The product of persulfide transfer to sulfite is thiosulfate, which is however not a substrate for persulfide dioxygenase (Kabil and Banerjee 2012). Of the persulfide donors that have been tested as substrates for persulfide dioxygenase only CoA-SSH exhibits activity in addition to GSSH (Kabil and Banerjee 2012). Hence, the mechanism by which the persulfide product of SQR is transferred to the persulfide dioxygenase and rhodanese is an important unanswered question in the field.