Three enzymatic activities catalyze the oxidation of sulfide to thiosulfate in mammalian and invertebrate mitochondria


T. Hildebrandt, Institut für Zoophysiologie, Heinrich-Heine-Universität Düsseldorf, 40225 Düsseldorf, Germany
Fax: +49 211 811 1676
Tel: +49 211 811 3577


Hydrogen sulfide is a potent toxin of aerobic respiration, but also has physiological functions as a signalling molecule and as a substrate for ATP production. A mitochondrial pathway catalyzing sulfide oxidation to thiosulfate in three consecutive reactions has been identified in rat liver as well as in the body-wall tissue of the lugworm, Arenicola marina. A membrane-bound sulfide : quinone oxidoreductase converts sulfide to persulfides and transfers the electrons to the ubiquinone pool. Subsequently, a putative sulfur dioxygenase in the mitochondrial matrix oxidizes one persulfide molecule to sulfite, consuming molecular oxygen. The final reaction is catalyzed by a sulfur transferase, which adds a second persulfide from the sulfide : quinone oxidoreductase to sulfite, resulting in the final product thiosulfate. This role in sulfide oxidation is an additional physiological function of the mitochondrial sulfur transferase, rhodanese.




glutathione persulfide


sulfide : quinone oxidoreductase

Hydrogen sulfide was known as a toxic pollutant long before its physiological functions became apparent. The main effects of sulfide poisoning are the loss of central respiratory drive due to lesions in the brain stem, and inhibition of cytochrome oxidase, leading to impaired aerobic energy metabolism [1]. Sulfide-rich environments occur naturally in the sediments and grass marshes of the intertidal zone, in deep-sea hydrothermal vents, and in the hypolimneon of eutrophic lakes [2]. Moreover, mammalian as well as invertebrate tissues, such as the body wall of the lugworm Arenicola marina, enzymatically produce sulfide [3,4].

In mammals, H2S acts as a gaseous transmitter and regulates several physiological processes [5]. Altered sulfide metabolism is associated with a number of disorders, such as Alzheimer’s disease, Down’s syndrome and ulcerative colitis [6–8]. Within a narrow concentration range, the effects of sulfide change from physiological to highly toxic, and therefore regulatory mechanisms are necessary to control endogenous sulfide levels within the physiological range.

Sulfide is catabolized mainly by oxidation to non-toxic sulfur compounds. Mammals rapidly oxidize sub-lethal concentrations of sulfide to sulfate and excrete it in the urine [1,9]. The main sites of sulfide oxidation are liver and colon tissues, and the enzyme activity is present in the mitochondria [10,11]. Thiosulfate is produced as an obligate intermediate by a putative sulfide oxidase, which has not yet been identified [12].

A mitochondrial pathway also catalyzes sulfide oxidation in many sulfide-adapted invertebrates including polychaetes, crustaceans and bivalves [2]. The lugworm A. marina is often used to study strategies of sulfide tolerance, because it is highly abundant in sandy to muddy intertidal flats, where high micromolar concentrations of sulfide regularly occur [2]. Mitochondrial sulfide oxidation in A. marina is linked to the electron transport chain at the level of ubiquinone via a sulfide : quinone oxidoreductase (SQR) [13]. As in most of the invertebrates studied so far, thiosulfate is the main excretion product of sulfide oxidation in A. marina [14], but the complete enzyme system catalyzing mitochondrial thiosulfate production is still unknown. In this paper, we provide further insights into the biochemical pathway of sulfide oxidation to thiosulfate in both mammalian and invertebrate mitochondria.


Mitochondrial sulfide oxidation

Mitochondria isolated from both rat liver and lugworm body-wall tissue quantitatively converted sulfide to thiosulfate, consuming corresponding amounts of molecular oxygen (Fig. 1A). The respiratory rates of isolated lugworm mitochondria were similar with added succinate (27.3 ± 3.5 nmol O2·mg−1·min−1) or sulfide (Table 1) as a substrate, but the respiratory control ratio was slightly higher with the carbon substrate (1.96 ± 0.41 versus 1.65 ± 0.17). In rat liver mitochondria, sulfide oxidation was significantly less coupled than succinate respiration (respiratory control ratios 1.41 ± 0.23 and 3.56 ± 0.98, respectively), with oxygen consumption rates for succinate respiration (58.0 ± 7.7 nmol O2·mg−1·min−1) fourfold higher than those for sulfide oxidation (Table 1). Myxothiazole and cyanide, which are inhibitors of complex III and complex IV of the respiratory chain, respectively, completely blocked sulfide oxidation.

Figure 1.

 Stoichiometries of mitochondrial sulfide oxidation and partial reactions in rat liver and in the body-wall musculature of Arenicola marina: decrease in substrate concentrations (white bars) and increase of product concentrations (grey bars) (μm). (A) Oxidation of 50 μm H2S by isolated mitochondria required oxygen, and thiosulfate was produced. (B) Mitochondrial membranes oxidized H2S to persulfides, which were detected as SCN after cyanolysis. Samples were taken after exactly 30 μm of the artificial electron acceptor decyl ubiquinone had been reduced. (C) Mitochondrial matrix fractions produced S2O32− from GSSH and O2. (D) Isolated sulfur dioxygenases oxidized GSSH to SO32−, consuming O2. (E) Purified sulfur transferases produced S2O32− from SO32− and GSSH. (F,G) Isolated sulfur dioxygenases mixed with purified persulfide transferase (F) or bovine rhodanese (G) converted GSSH and O2 to S2O32−. Samples were taken after 79.3 ± 4.5 μm GSSH had been consumed.

Table 1.   Rates of sulfide and persulfide oxidation. Rates of H2S oxidation in isolated rat and lugworm mitochondria (nmol O2·mg−1·min−1), sulfide : quinone oxidoreductase activities of mitochondrial membranes (nmol reduced decyl ubiquinone mg−1·min−1), and rates of GSSH oxidation in mitochondrial matrix fractions (nmol O2·mg−1·min−1).
CompartmentActivity of the pathway
Mitochondria14.39 ± 2.3725.43 ± 2.22
Membranes56.63 ± 8.21226.32 ± 11.41
Matrix138.11 ± 27.95133.16 ± 14.19

Sulfide : quinone oxidoreductase

Membrane preparations from mitochondria of rat liver as well as lugworm body-wall tissue were able to reduce externally added decyl ubiquinone in the presence of sulfide (100 μm) as a substrate. When 30 μm decyl ubiquinone was reduced, approximately the same concentration of sulfide was consumed, and stoichiometric amounts of cyanolysable sulfane sulfur (detected as SCN) were produced (Fig. 1B). Neither thiosulfate, sulfite or sulfate accumulated in the reaction mixture. SQR activity was four times higher in lugworm than in rat mitochondrial membranes (Table 1). Substrate saturation was achieved at low micromolar sulfide concentrations (Km = 9.94 ± 1.36 μm for lugworm and 2.87 ± 0.29 μm for rat membranes). The SQR activity of rat mitochondrial membranes decreased when the ambient sulfide concentration exceeded 300 μm, whereas SQR activity of lugworm mitochondrial membranes was not inhibited by up to 2 mm H2S. Hardly any SQR activity was detectable if KCN was omitted from the assay. Thus, the persulfide acceptor is essential for the SQR reaction in vitro.

Persulfide oxidation in the mitochondrial matrix

SQR produces sulfane sulfur, so the complete reaction of sulfide to thiosulfate requires at least one further oxidative step. Therefore, we looked for an enzyme activity capable of elemental sulfur oxidation. In combination with 1 mm glutathione (GSH), elemental sulfur was oxidized to thiosulfate in an oxygen-dependent manner by isolated matrix fractions of rat and lugworm mitochondria (Fig. 2A,B). The sulfane sulfur-oxidizing enzyme activity is tentatively referred to as sulfur dioxygenase, as it corresponds to an enzyme catalyzing the glutathione-dependent oxidation of elemental sulfur in acidophilic thiobacilli, in which glutathione persulfide (GSSH) is the actual substrate [15]. GSSH was generated by injection of a saturated acetonic sulfur solution into the GSH-containing assay mixture. In the rat and lugworm mitochondrial matrix, half maximal sulfur dioxygenase activities were achieved in the presence of 0.31 ± 0.03 and 0.22 ± 0.03 mm GSH, respectively. Neither elemental sulfur nor GSH were metabolized when added separately. The acetone used as a solvent for elemental sulfur did not inhibit sulfur dioxygenase activity.

Figure 2.

 Original traces of persulfide oxidation and partial reactions in the mitochondrial matrix of rat liver and the body-wall musculature of Arenicola marina. Time courses of complete reactions of 80 μm GSSH (bsl00041) were recorded, together with the corresponding concentrations of O2 (—), S2O32− (bsl00000) and SO32− (bsl00066). (A) 65 μg·mL−1 rat mitochondrial matrix; (B) 48 μg·mL−1 lugworm mitochondrial matrix; (C) 20 μg·mL−1 partially purified rat sulfur dioxygenase; (D) 9 μg·mL−1 partially purified lugworm sulfur dioxygenase; (E) 3 μg·mL−1 purified rat sulfur transferase; (F) 2 μg·mL−1 partially purified lugworm sulfur transferase.

Persulfide oxidation rates were identical in the rat and lugworm mitochondrial matrix (Table 1). Complete oxidation of the 79.3 ± 4.5 μm GSSH that was used as a substrate required about 40–50 μm molecular oxygen, with thiosulfate being the main product (Fig. 1C). Only small amounts of sulfite and sulfate (≤ 5 μm) accumulated. No oxygen consumption or thiosulfate production was detectable with sulfide as a substrate.

Two enzymes were necessary to convert sulfane sulfur to thiosulfate in the mitochondrial matrix. The sulfur dioxygenase produced sulfite and only minor amounts of thiosulfate from GSSH and molecular oxygen (Figs 1D and 2C,D). No further cofactors were necessary, but the sulfur dioxygenase activity increased by about 25% in the presence of ascorbate (2.5 mm). The sulfur dioxygenase could be purified 44-fold from rat liver and 72-fold from lugworm body-wall tissue, resulting in specific activities of 0.87 ± 0.04 and 0.85 ± 0.24 U·mg−1, respectively. Based on the results of size-exclusion chromatography, both enzymes had a molecular weight of about 46 kDa. Unfortunately the yields were rather low (3–7%), and homogeneity was not achieved as the sulfur dioxygenase activity rapidly disappeared after ion chromatography.

The final reaction step was catalyzed by a sulfur transferase, which is defined as an enzyme that transfers sulfane atoms from a donor molecule to a thiophilic acceptor substrate [16]. The sulfur transferases isolated from both rat and lugworm mitochondria stoichiometrically transferred persulfide groups from GSSH to sulfite, producing thiosulfate (Figs 1E and 2E,F), and this activity is referred to as persulfide transferase activity. The functional role of the sulfur transferase for thiosulfate production during persulfide oxidation was demonstrated by size-exclusion chromatography of the mitochondrial matrix constituents (Fig. 3). Only if the sulfur dioxygenase-containing fractions (peak fractions from 55.5–57.5 mL) were combined with specific later-eluting fractions (63–69 mL for rat and 69–75 mL for lugworm samples) could GSSH be converted to thiosulfate (Fig. 3, open triangles). The main mitochondrial sulfur transferase, rhodanese, is normally detected by its in vitro activity as a thiosulfate : cyanide sulfur transferase on the basis of thiocyanate production [17]. For both animals studied, rhodanese activity (Fig. 3, filled triangles) co-eluted with the thiosulfate-producing enzyme, which was purified to homogeneity from rat liver. It comprised a single polypeptide of about 35 kDa (Fig. 4A), and the sequence of tryptic peptides was identical with the published rhodanese sequence (EC [18]. The final purification product of the lugworm sulfur transferase was enriched 406-fold, but still contained myoglobin and two proteins of approximately 16.5 kDa that could not be identified (Fig. 4B).

Figure 3.

 Size-exclusion chromatography (Superdex 75) of mitochondrial matrix fractions from (A) rat liver and (B) the body-wall musculature of Arenicola marina. Conditions: 0.1 m phosphate buffer (pH 7.4), flow rate=0.5 mL·min−1, fraction volume = 2 mL. Volume activities (percentage of maximal activity) were determined for sulfur dioxygenase (bsl00041) and rhodanese (bsl00066). In addition, the concentrations (μm) of SO32− (bsl00042) and S2O32− (Δ) (the products of persulfide oxidation) in each fraction mixed with the same volume of fraction 28 (55.5–57.5 mL elution volume) are shown. Gray lines indicate the absorbance at 280 nm.

Figure 4.

 Silver-stained SDS–PAGE gels of purified sulfur transferases. m, molecular mass standard (size indicated). (A) Sulfur transferase from rat liver mitochondrial matrix after size-exclusion and cation-exchange chromatography (4 μg). (B) Sulfur transferase from the matrix fraction of mitochondria isolated from the body-wall musculature of A. marina after size-exclusion chromatography (5.5 μg).

When mixed with the isolated sulfur transferases, the sulfur dioxygenases from rat and lugworm mitochondria oxidized persulfides to thiosulfate, showing a stoichiometry similar to that for the matrix fractions (Fig. 1F). The sulfur transferase could be functionally replaced by bovine rhodanese purchased from Sigma-Aldrich (Taufkirchen, Germany) (Fig. 1G).

The affinities of the sulfur transferases from rat and bovine liver, as well as the lugworm enzyme, for the substrates of the persulfide transferase reaction, persulfides and sulfite, were significantly higher than for the substrates of the rhodanese reaction, thiosulfate and cyanide (Table 2). The Km values for GSSH and sulfite were in the low micromolar range, whereas millimolar concentrations of thiosulfate were required for half-maximal rhodanese activity. Differences were seen with respect to the enzyme activities at substrate saturation (Table 2). The rat liver sulfur transferase produced thiocyanate about 150 times faster than thiosulfate, and for bovine rhodanese the ratio was 230. In contrast, the lugworm sulfur transferase had more than twice as much persulfide transferase activity as rhodanese activity.

Table 2.   Kinetic properties of the persulfide transferase and rhodanese activities of sulfur transferases. Persulfide transferase activities (1 unit = 1 μmol S2O32−·min−1) and rhodanese activities (1 unit = 1 μmol SCN·min−1) of sulfur transferases purified from rat liver and the body-wall musculature of Arenicola marina and of bovine liver rhodanese purchased from Sigma-Aldrich. The Km values were determined by fitting the data to the Michaelis–Menten equation using sigmaplot version 9.01 (Systat Software) and the enzyme kinetic module 2.0.
 Persulfide sulfur transferaseThiosulfate sulfur transferase (rhodanese)
Activity (U·mg−1)Km GSSH (μm)Km SO32−m)Activity (U·mg−1)Km S2O32− (mm)Km KCN (mm)
Rat4.15 ± 0.5257.3 ± 7.17.93 ± 0.39615.2 ± 106.912.4 ± 1.60.31 ± 0.02
Lugworm8.52 ± 1.0070.0 ± 20.246.3 ± 5.03.29 ± 0.464.32 ± 0.267.51 ± 0.52
Bovine0.40 ± 0.1161.3 ± 17.821.8 ± 3.694.5 ± 5.616.2 ± 2.00.087 ± 0.009


Three enzymes are required for the production of thiosulfate from hydrogen sulfide in the mitochondria of a sulfide-adapted invertebrate, the lugworm Arenicola marina, as well as in vertebrate rat liver mitochondria (Fig. 5). The first step is catalyzed by a membrane-bound sulfide : quinone oxidoreductase, which oxidizes sulfide to sulfane sulfur, transferring two electrons to the ubiquinone pool. A putative sulfur dioxygenase in the mitochondrial matrix requires molecular oxygen and water for the subsequent four-electron oxidation of one persulfide molecule to sulfite. A second persulfide is added by a sulfur transferase, resulting in the final product thiosulfate. In rat liver mitochondria, thiosulfate can be further metabolized to sulfate, probably by thiosulfate reductase and sulfite oxidase [10,19].

Figure 5.

 Proposed model of mitochondrial sulfide oxidation. A membrane-bound sulfide : quinone oxidoreductase (SQR) oxidizes sulfide (H2S) to the level of elemental sulfur, simultaneously reducing a cysteine disulfide such that a persulfide group is formed at one of the cysteines (SQR-SSH) [13]. The electrons are fed into the respiratory chain via the quinone pool (Qox/Qred), and finally transferred to oxygen by cytochrome oxidase (complex IV). A sulfur dioxygenase in the mitochondrial matrix oxidizes persulfides to sulfite (H2SO3), consuming molecular oxygen and water. The final reaction is catalyzed by a sulfur transferase, which produces thiosulfate (H2S2O3) by transferring a second persulfide from the SQR to sulfite.

First step: sulfide : quinone oxidoreductase

The SQR from A. marina has recently been heterologously expressed in yeast and functionally characterized [13]. Similar genes were also detected in the human and mouse genomes [20]. The results of the present study confirm the functional relevance of the SQR activity in invertebrate and mammalian mitochondria. The complete inhibition of sulfide oxidation by myxothiazole demonstrates that sulfide is exclusively oxidized via the SQR pathway in rat liver mitochondria. Similar results have been obtained for A. marina [21].

Isolated mitochondrial membranes converted sulfide to cyanolysable sulfane sulfur, reducing equimolar concentrations of decyl ubiquinone. This reaction is in accordance with a two-electron oxidation of each sulfide molecule to the level of elemental sulfur during the first oxidative step, which is linked to the respiratory chain via ubiquinone. Thus, the electrons can be used for chemolithotrophic ATP production, as demonstrated for several sulfide-adapted animals such as A. marina and the mussel Geukensia demissa, and also indirectly for mammalian cells [21–23]. The respiratory control ratios > 1 obtained in this study indicate coupling of sulfide oxidation with oxidative phosphorylation in rat liver mitochondria.

The low Km values of rat and lugworm SQR for sulfide are in the same range as reported for bacterial SQRs (2–26 μm sulfide) [24], and allow maximal oxidation rates at physiological sulfide concentrations. The primary oxidation product is presumably a persulfide bound to a cysteine residue of the SQR [13]. In vivo, the enzymes catalyzing the two subsequent reactions of the pathway may function as a persulfide acceptor instead of cyanide. Rhodanese interacts with various enzymes and is partially bound to the inner mitochondrial membrane [25].

Second step: sulfur dioxygenase

Glutathione persulfides are oxidized to thiosulfate in the mitochondrial matrix in an oxygen-dependent manner. This reaction is identical with sulfur dioxygenase activities, which have been described for the sulfur bacteria Acidothiobacillus thiooxidans, Acidothiobacillus ferrooxidans and Acidiphilium acidiphilum [26]. The activity assay requires GSH as a catalyst for elemental sulfur activation [15]. The Km values of the mitochondrial sulfur dioxygenases for GSH are similar to those calculated for the thiobacilli (120–240 μm GSH) [15]. In vivo, the sulfur dioxygenase very probably uses internally produced persulfides as a substrate, e.g. those resulting from the SQR reaction. Small thiols such as GSH and dihydrolipoate or thioredoxin could be involved in the transfer of persulfide groups between the enzymes of the pathway [27,28]. For each molecule of O2 consumed, approximately two molecules of GSSH were converted to one thiosulfate molecule in the mitochondrial matrix. This ratio corresponds to the four-electron transfer necessary to complete sulfide oxidation. However, thiosulfate production did not exactly match oxygen consumption, most notably in rat matrix preparations. Dilution of the matrix enzymes probably promoted side reactions by limiting substrate channelling between sulfur dioxygenase and sulfur transferase. Thiosulfate production increased when both enzymes were purified separately and mixed afterwards (Fig. 1F), such that interfering matrix components such as transition metals catalyzing auto-oxidation of persulfides and sulfite or carbonyl and thiol compounds, as well as nicotinamide, flavin and folate derivatives, which bind sulfite, were removed [29,30].

The partially purified mitochondrial sulfur dioxygenases produced sulfite, albeit less than would be expected when compared to the concentration of GSSH consumed. One possible explanation for this discrepancy may be that product inhibition of the sulfur dioxygenase promotes non-enzymatic reactions of excess GSSH [15]. Sulfite is highly reactive and cytotoxic mainly due to increased formation of reactive oxygen species [31]. It is also known to produce protein S-sulfonates and polysulfane monosulfanic acid, which would lower the free sulfite concentration as well [32,33].

Apart from molecular oxygen, no additional electron acceptors were necessary. The sulfur dioxygenase might therefore resemble cysteine dioxygenase, a cytosolic non-heme iron dioxygenase, which also catalyzes the oxidation of a sulfhydryl group by incorporation of molecular oxygen [34]. The experimental loss of Fe2+ during ion-exchange chromatography would explain the rapid inactivation of the mitochondrial sulfur dioxygenases. Similar problems have been reported during the purification of bacterial sulfur dioxygenases and rat cysteine dioxygenase [35,36]. The activating effect of ascorbate is also consistent with the postulated mechanism, as other dioxygenases such as prolin hydroxylases also require vitamin C to retain full activity [37]. Further experiments will be necessary to achieve complete purification and identify the sequence of the sulfane sulfur-oxidizing enzyme, before final evidence for the proposed reaction mechanism can be provided.

Third step: sulfur transferase

A sulfur transferase is essential for the production of thiosulfate from sulfane sulfur in the mitochondrial matrix. Therefore, the present results disagree with the non-enzymatic reaction of the primary product sulfite with GSSH suggested for the partially purified bacterial sulfur dioxygenase [35]. The persulfide transferase activity is a physiological function of the mitochondrial sulfur transferase rhodanese, thus proving the in vivo relevance of persulfides and sulfite, both of which have been suggested as possible substrates [38]. The double displacement mechanism (Fig. 6), which has been postulated for the rhodanese reaction [39], can also explain the persulfide transferase activity of the sulfur transferase. The SQR serves as a persulfide donor and transfers the sulfane atom produced during the first step of sulfide oxidation to an active site cysteine residue of the sulfur transferase. In vitro, GSSH provides a persulfide group instead of the SQR. Subsequently, sulfite reacts with the covalent enzyme–sulfur intermediate to produce thiosulfate.

Figure 6.

 Reaction mechanism of a sulfur transferase with (A) rhodanese activity and (B) persulfide transferase activity. (A) Rhodanese mechanism as described by Westley [39]: in a double-displacement mechanism, a sulfur atom is transferred from thiosulfate to an active site cysteine thiol. The first product, sulfite, is released before cyanide reacts with the enzyme-bound persulfide, forming the second product, thiocyanate. (B) The same double-displacement mechanism explains the persulfide transferase activity of the sulfur transferase, for which glutathione persulfide is the persulfide donor and sulfite is the persulfide acceptor.

Although rhodanese has been known for 75 years [40] and is widely distributed among prokaryotes and eukaryotes [39], its biological role is still a matter of debate. Proposed functions include cyanide detoxification, maintenance of the sulfane pool [16], formation of iron–sulfur clusters and regulation of oxidative phosphorylation [25]. Furthermore, several possible roles in sulfide oxidation have been suggested [38,41–44]. The Km values for the rhodanese activity of the purified sulfur transferases from rat and lugworm reported here are comparable to kinetic data obtained previously [45,46], and the low substrate affinities strongly suggest that cyanide and thiosulfate are not the natural substrates of the sulfur transferase. In constrast, the Km values for sulfite and persulfides, the substrates of the persulfide transferase reaction, are all within the physiological range.

The detrimental effects of the rhodanese deficiency found in colonocytes of patients with ulcerative colitis and colorectal cancer [47] can be explained if the sulfur transferase is part of the sulfide oxidation pathway. The elevated intestinal sulfide concentrations associated with these diseases possibly result from an accumulation of toxic sulfite, which inhibits the previous enzymatic steps of the pathway.

Biological implications of the sulfide oxidation pathway

The mitochondrial pathway catalyzing sulfide oxidation in three consecutive steps can be integrated into the cellular metabolism at the level of each intermediate. Persulfides are produced by the SQR, eliminated by the sulfur dioxygenase and distributed between the various acceptors by sulfur transferases.

For intertidal invertebrates, which often experience increasing sulfide concentrations in combination with hypoxia, this sulfide oxidation pathway is particularly favourable, as the oxygen demand can be reduced by arresting sulfide oxidation after the SQR reaction at the level of sulfane sulfur. This strategy has already been demonstrated for the sipunculid worm Phascolosoma arcuatum, for the mudskipper Boleophthalmus boddaerti and for the nematode Oncholaimus campylocercoides, all of which store or excrete products in the oxidation state of elemental sulfur during hypoxia [44,48,49].

The same metabolic pathway catalyzes mitochondrial sulfide oxidation in rat and lugworm, i.e. mammalian and invertebrate mitochondria. Therefore, the various strategies of sulfide detoxification, which have been partially unravelled for various sulfide-adapted species, could also be based on the same principle. Sulfite and sulfane sulfur have already been identified as intermediates of mitochondrial sulfide oxidation for the mussel Solemya reidi [42].

In A. marina, the kinetic properties of the individual enzymes reveal an adaptive strategy towards effective oxidation of high sulfide concentrations. The SQR activity is four times higher and markedly more sulfide-tolerant in lugworm than in rat mitochondrial membranes, and persulfide transferase activity is dominant in the lugworm sulfur transferase, in contrast to the rat enzyme, which is presumably equally engaged in several metabolic functions in addition to sulfide oxidation.

Experimental procedures


Experiments were carried out in accordance with the European Communities Council Directive (86/609/EEC) regarding the care and use of animals for experimental procedures. Livers were taken from adult Wistar rats that had been fed ad libitum and were killed by carbon dioxide asphyxiation. Specimens of Arenicola marina from the intertidal flats near Texel, The Netherlands, were purchased from Zeeaashandel Arenicola (in Texel).

Purification procedures

Rat livers were homogenized in ice-cold isolation medium (250 mm sucrose, 10 mm triethanolamine, 1 mm EGTA, 2 mm KH2PO4 and 2 mm K2HPO4, pH 7.4), and centrifuged at 600 g for 10 min followed by 9000 g for 20 min. The mitochondrial pellet was resuspended in incubation medium (250 mm sucrose, 10 mm triethanolamine, 1 mm EGTA, 2 mm KH2PO4, 2 mm K2HPO4 and 5 mm MgCl2, pH 7.4). Mitochondria were isolated from the body-wall musculature of A. marina as described by Völkel and Grieshaber [21]. Mitochondrial membranes and matrix fractions were prepared according to the method described by Lass and Sohal [50], except that the mitochondrial pellets were resuspended in 0.1 m potassium phosphate buffer, pH 7.4.

Matrix fractions were applied to a Superdex 75 column (HiLoad 16/60, GE Healthcare, Freiburg, Germany) and eluted in 0.1 m phosphate buffer, pH 7.4. Fractions with maximal sulfur dioxygenase or persulfide transferase activity, respectively, were pooled and the proteins were concentrated using Amicon Ultra centrifugal devices (Millipore, Eschborn, Germany). Sulfur dioxygenase from rat liver was further purified by anion-exchange chromatography on a Resource Q column (1 mL, GE Healthcare) equilibrated with 20 mm Tris, pH 8.0, using a linear gradient from 70 to 150 mm NaCl in 30 mL. Final purification of rat liver persulfide transferase was achieved on a Source S column (1 mL, GE Healthcare) equilibrated with 50 mm Mes, pH 6.0, with a linear gradient from 0 to 100 mm NaCl in 20 mL. Partially purified sulfur dioxygenase and persulfide transferase from A. marina (fractions from the Superdex 75 column) were used for the activity assays.

Protein content was measured according to the Bradford method [51] using BSA (fraction V, Sigma-Aldrich) as a protein standard. The purified enzyme fractions were analyzed on 12% SDS–PAGE gels. Some protein spots were subjected to a tryptic in-gel digestion, purified and desalted (Perfect Pure C18, Eppendorf, Hamburg, Germany) and analysed by ESI-MS using an ESI-QqTOF (QSTAR XL, Applied Biosystems, Darmstadt, Germany) equipped with a nanospray ion source.

Assays of mitochondrial respiration

Mitochondrial oxygen consumption was measured essentially as described by Völkel and Grieshaber [21] using an oxygraph respirometer (Oroboros, Innsbruck, Austria) at 25 °C. Myxothiazole (5 μm, dissolved in ethanol) and KCN (0.5 mm, dissolved in water) were used to inhibit complex III and complex IV of the respiratory chain, respectively.

Enzyme assays

All enzyme assays were performed at 25 °C in air-saturated solutions.

SQR activity of mitochondrial membranes was determined by following the enzymatic reduction rate of decyl ubiquinone (Sigma-Aldrich) at 275 nm upon sulfide addition (100 μm) [13].

The sulfur dioxygenase activity (1 unit = 1 μmol O2·min−1) was measured polarographically in an oxygraph respirometer (Oroboros) using a modification of the assay described by Suzuki [52]. The reaction mixture (2 mL) consisted of 0.1 m potassium phosphate buffer (pH 7.4), 1 mm GSH and 2–10 μg·mL−1 of the partially purified enzyme. The reaction was started by injecting 30 μL of a saturated acetonic sulfur solution (see below), resulting in a final concentration of 79.3 ± 4.5 μm GSSH in the oxygraph chamber. Oxygen consumption was calculated based on an initial oxygen concentration in air-saturated phosphate buffer of 244 μm [53].

The enzymatic production of thiosulfate from glutathione persulfide and sulfite is referred to here as persulfide transferase activity. The reaction mixture contained 0.1 m potassium phosphate buffer (pH 7.4), 1 mm GSH and sulfur transferase. Simultaneous addition of 15 μL·mL−1 of a saturated acetonic sulfur solution and 200 μm Na2SO3 initiated the reaction. The thiosulfate concentration was determined after appropriate time intervals (1–30 min). As the rhodanese preparation purchased from Sigma-Aldrich contained about 1 mm thiosulfate, it was applied to a PD-10 column (GE Healthcare) and transferred into 0.1 m potassium phosphate buffer (pH 7.4) before use.

Rhodanese activity was determined by measuring the formation of thiocyanate from cyanide and thiosulfate [17].

The apparent Km values for the persulfide transferase and rhodanese activities of the sulfur transferases were determined at the optimal protein concentrations and reaction times resulting in constant rates of product formation. Kinetic parameters were calculated by non-linear least square analysis of the data fitted to the Michaelis–Menten equation using the enzyme kinetics module of sigmaplot version 9.01 (Systat Software, Erkrath, Germany).

Substrate solutions

A saturated acetonic sulfur solution stored on ice was used as a stock solution for the assays of enzymatic sulfur oxidation and sulfane sulfur transfer. To check whether the acetone had any effect on the enzyme activity, a solution containing dispersed sulfur in water was prepared according to the method described by Rohwerder and Sand [15].

Na2S and Na2SO3 stock solutions were prepared immediately before use in nitrogen-saturated distilled water.

Determination of sulfur compounds

Dissolved sulfide was analysed photometrically using a commercial sulfide test (Spektroquant, Merck, Darmstadt, Germany). Samples were fixed with 50 mm zinc acetate in 150 mm NaOH prior to analysis.

The initial concentration of GSSH in the assays routinely used for sulfur dioxygenase and persulfide transferase activity tests was determined by cold cyanolysis [54].

The ability of persulfides to reduce methylene blue under acidic conditions was used to measure the rates of substrate consumption during sulfane sulfur oxidation and transfer in the mitochondrial matrix. Samples (250 μL) were taken from the sulfur dioxygenase and persulfide transferase activity assays at various time intervals and incubated with 375 μL of 10 m HCl and 375 μL of 75 μm methylene blue at room temperature for 30 min. The absorption of oxidized methylene blue was determined at 670 nm, and subtracted from that of blanks containing buffer instead of the sample. As methylene blue is also reduced by thiosulfate, separate calibration curves were produced for gluthatione persulfide and thiosulfate, and each data point was corrected for the actual thiosulfate concentration in the sample as estimated by HPLC (see below).

Thiocyanate was detected photometrically [17].

The concentrations of sulfite and thiosulfate were measured using the monobromobimane HPLC method, and sulfate was quantified by high-performance ion chromatography [14].

Controls were performed using heat-inactivated mitochondria, membranes and enzyme preparations (30 min at 85 °C).

Statistical analyses

Data are given as means ± standard deviation of the results from 3–12 preparations. Significant differences between means were evaluated by t-tests at the < 0.05 level using a statistical software package (sigmastat version 3.1; Systat Software).


We thank Barbara Bartosinska and Silke Jakob (Institut für Zoophysiologie, Heinrich-Heine-Universität Düsseldorf, Germany) for skilful technical assistance, and Ursula Theissen for valuable discussions (Institut für Botanik, Heinrich-Heine-Universität Düsseldorf, Germany). This work was supported by the Deutsche Forschungsgemeinschaft (GR 456/22-1).