Hypotaurine and sulfhydryl-containing antioxidants reduce H₂S toxicity in erythrocytes from a marine invertebrate

A variety of marine animals live in habitats such as deep-sea hydrothermal vents, hydrocarbon seeps, mudflats and estuaries where they are exposed to the toxin sulfide (representing the sum of H₂S, HS^–and S^2–) either continuously or periodically(Gamenick et al., 1998). In addition to inhibiting cytochrome c oxidase(Dorman et al., 2002; Khan et al., 1990; Nicholls and Kim, 1982) and a variety of other enzymes (Bagarinao,1992), sulfide may increase oxidative damage via the generation of free radicals. Chen and Morris first recognized that the oxidation of sulfide would result in a chain of redox reactions, producing a variety of free radical intermediates, including reactive oxygen species and sulfur-centered free radicals (Chen and Morris, 1972). This was later confirmed experimentally in seawater(Tapley et al., 1999) and in animal tissues in vitro (Tapley,1993; Truong et al.,2006). However, sulfide may also have antioxidant properties:H₂S can directly scavenge reactive oxygen and nitrogen species(Whiteman et al., 2004; Whiteman et al., 2005), and reduce lipid peroxidation in a mouse model of hepatic ischemia-reperfusion injury (Jha et al., 2008). Nonetheless, oxidative stress is increased in at least some marine organisms during sulfide exposure (Abele-Oeschger and Oeschger, 1995; Joyner-Matos et al., 2007; Julian et al.,2005; Tapley,1993), as well as in mammalian cells exposed to sulfide in vitro (Attene-Ramos et al.,2007; Eghbal et al.,2004). Furthermore, environmental conditions other than sulfide,in particular high concentrations of dissolved metals, may contribute to oxidative stress at hydrothermal vents(Bebianno et al., 2005). Animals in sulfidic habitats appear to minimize the toxic effects of sulfide primarily by reducing the free sulfide concentration, especially by enzymatic oxidation of sulfide to thiosulfate and by covalent binding of sulfide with sulfhydryl groups of `sulfide-binding proteins'(Arp and Childress, 1983; Arp et al., 1984; Arp et al., 1987; Grieshaber and Völkel,1998; O'Brien and Vetter,1990).

Hypotaurine (HT), the reduced metabolic precursor of taurine, is an organic osmolyte that is present in high concentrations (as much as 150 mmol kg^–1 tissue wet mass) within the hemolymph and cells of many marine invertebrates at deep-sea hydrothermal vents and hydrocarbon seeps(Alberic, 1986; Horak et al., 2003; Pruski et al., 2000a; Rosenberg et al., 2006; Yancey, 2005; Yancey et al., 2002b; Yin et al., 2000). HT is an effective hydroxyl radical scavenger in mammalian cells in vitro(Aruoma et al., 1988), and HT and its precursors, including cysteamine, cysteinesulphinic acid and cysteic acid, have been proposed to act as antioxidants in mammalian brain, male reproductive tissue and liver cells (Aruoma et al., 1988; Donnelly et al.,2000; Fellman and Roth,1985; Huxtable,1992; Schurr and Rigor,1987). Therefore, HT may reduce sulfide toxicity by scavenging free radicals produced from sulfide oxidation. However, HT can also react with sulfide directly, forming thiotaurine (ThT) in vitro(Cavallini et al., 1963). If this reaction occurs in vivo, it would scavenge free sulfide and might, therefore, reduce sulfide toxicity by acting as a sulfide `buffer'(Alberic and Boulegue, 1990; Brand et al., 2007; Joyner et al., 2003; Pruski and Fiala-Medioni,2003; Rosenberg et al.,2006). In sulfide-exposed animals that lack sulfide-oxidizing endosymbionts, the resulting ThT may be enzymatically recycled back to HT,thereby releasing sulfide at a rate that allows detoxification through other mechanisms (Rosenberg et al.,2006). In sulfide-exposed animals with sulfide-oxidizing endosymbiotic bacteria, the conversion of HT to ThT may provide a mechanism to transport sulfide to the endosymbionts, which would then convert ThT back to sulfide and HT (Pranal et al.,1995; Pruski and Fiala-Medioni, 2003; Pruski et al., 1997; Pruski et al.,2001; Pruski et al.,2000b). Accordingly, Pranal and colleagues proposed that the ratio of ThT to HT in tissues represents the extent of that animal's recent sulfide exposure (Pranal et al.,1995). This was confirmed by Brand and colleagues and Rosenberg and colleagues for vent species with and without sulfide-oxidizing endosymbionts, respectively (Brand et al.,2007; Rosenberg et al.,2006). Therefore, HT may have two beneficial actions in sulfide-exposed animals: firstly, scavenging free radicals generated by the oxidation of sulfide and secondly, scavenging sulfide by reacting with it to form ThT (Joyner et al., 2003; Pruski et al., 2000b; Yancey et al., 2002b). However, whether HT actually reduces sulfide toxicity and, if so, whether this is primarily achieved by scavenging free radicals or by scavenging sulfide have never been directly tested.

In the present study, we investigated whether HT, ThT, antioxidants containing sulfhydryl groups and antioxidants without sulfhydryl groups can reduce the cytotoxicity of sulfide exposure in vitro. Sulfide cytotoxicity was assessed by exposing erythrocytes from the bloodworm Glycera dibranchiata Ehlers 1868 (Annelida: Polychaeta: Phyllodocida:Glyceridae) to sulfide for 2 h, followed by the application of vital fluorescent dyes to determine cell viability as indicated by plasma membrane integrity. G. dibranchiata inhabits mudflats along the North Atlantic coast of the United States, where it is probably exposed to sulfide. It lacks a vascular system and its coelomic, hemoglobin-containing erythrocytes are circulated to the parapodia for gas exchange(Mangum, 1994) and, therefore,are at risk for sulfide exposure. Body wall and coelomocyte extracts from G. dibranchiata have been reported to contain HT but the concentrations were not determined (Costa et al., 1980). However, Yancey indicated that Glycera sp. are among the invertebrates for which HT is not a major organic osmolyte in muscle or whole-body tissue (Yancey,2005; Yancey et al.,2002b). This suggests that if G. dibranchiata tissues contain endogenous HT, the concentrations are probably comparatively low. Consequently, an effect of exogenous HT on G. dibranchiata cell viability would presumably be independent of any physiological adaptations that might otherwise utilize endogenous HT.

The following seven compounds and antioxidants were tested for their ability to reduce the dissolved sulfide concentration and increase cell viability during H₂S exposure: (1) HT (0.5, 5.0, 50 mmol l^–1); (2) ThT (0.5, 5.0 mmol l^–1); (3)glutathione ethyl ester (GEE, 0.03, 0.3, 3 mmol l^–1), a cell-permeant analogue of glutathione that is commonly used at 0.01–2 mmol l^–1in vitro (e.g. Doolen et al., 2007; Pocernich et al., 2001); (4)N-acetylcysteine (NAC, 0.03, 0.3, 1.0 mmol l^–1), a cell-permeant N-acetyl derivative of L-cysteine that is a glutathione precursor and is commonly used at 0.5–20 mmol l^–1in vitro (e.g. Konarkowska et al., 2005; Yan et al.,2006); (5) l-ascorbic acid (ASC, 0.001, 0.01, 0.1 mmol l^–1), i.e. vitamin C, which is cell-impermeant and is commonly used at 0.001–0.1 mmol l^–1in vitro(e.g. Yedjou et al., 2008);(6) 4-hydroxy-2,2,6,6-tetramethylpiperidine-1-oxyl (Tempol, 0.03, 0.3, 3.0 mmol l^–1), a cell-permeant nitroxide antioxidant that scavenges superoxide radicals and is commonly used at 0.003–10 mmol l^–1in vitro (e.g. Samuni et al., 2004; Singh et al., 2007); and (7)6-hydoxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox®, 0.001,0.01, 0.1 mmol l^–1 from a 100 mmol l^–1 stock solution in ethanol), a water-soluble, cell-permeant analog ofα-tocopherol that is commonly used at 0.002–100 mmol l^–1in vitro with mammalian cells (e.g. Bhattacharya and Lakshmana Rao,2001; Casado et al.,2007) and mussel hemocytes(Machella et al., 2006),although its aqueous solubility is limited above 1.8 mmol l^–1(Wu et al., 1990). All compounds except ThT were obtained from Sigma Chemical Co. (St Louis, MO, USA)and were the highest quality available. ThT was prepared by reacting HT with sulfide as described previously (Cavallini et al., 1963; Joyner et al.,2003).

G. dibranchiata were purchased from Harbor Bait Company,Wiscasset, ME, USA, which collected the worms from mudflats during low tide. The worms were maintained in the laboratory unfed in filtered, 15°C chilled seawater for no longer than two weeks before being used. Worms were removed from the aquarium immediately prior to experiments, and the coelomic fluid was drained from an incision in the proboscis and then centrifuged for 18 s at 12,000× g in order to remove the supernatant and overlying white cell layer. The pelleted erythrocytes were further purified by centrifugation for 10 min at 170×g and 10°C into a 25% sucrose–seawater cushion. The erythrocytes were then washed with cold incubation buffer (sterile-filtered seawater with 10 mmol l^–1 Hepes and 0.1% glucose, pH 7.6, 1000 mosmol kg^–1) to remove sucrose residue, and the pelleted erythrocytes were then diluted 140-fold in cold incubation buffer.

Sulfide exposures were conducted by exposing a thin layer of incubation buffer, either with or without erythrocytes, to a mixture of H₂S gas in air. This was performed in five gas-tight chambers, each containing a black-wall, clear-bottom 96-well microplate (Corning Life Sciences, Lowell,MA, USA), in which wells contained 50 μl of incubation buffer. An appropriate volume of H₂S gas (from a compressed tank of 99% pure H₂S gas) was then added by syringe to the gas space of each chamber to obtain 0 (control), 0.10, 0.32, 1.0 and 3.2% H₂S in air. After exposure, the sulfide concentration in each well was determined with a Methylene Blue assay (Gilboa-Garber,1971) modified for 50 μl sample volumes.

To determine the effect of HT and the antioxidants on the dissolved sulfide concentration, 50μl of incubation buffer containing HT (0.5, 5.0, 50 mmol l^–1) or the highest concentrations of GEE, NAC, ASC, Tempol and Trolox were exposed to H₂S for 2 h at 22–24°C, after which the dissolved sulfide concentration was determined, all as described above (note that ThT was not tested in this experiment due to the expense of synthesizing it, given the low probability that it would have any effect on dissolved sulfide).

To determine the effect of sulfide on cell viability, 50 μl of erythrocytes (diluted 1:1400 in incubation buffer) was added to microplate wells and exposed to H₂S gas, as described above, for 2 h at 15°C. This experiment was independently replicated with erythrocytes from three worms, with erythrocyte samples from each worm being exposed to control conditions and each level of H₂S (i.e. a fully balanced design). At each H₂S concentration for each replicate, an additional well containing 50 μl of incubation buffer without erythrocytes was used for the determination of the dissolved sulfide concentration, as described above.

To determine whether the addition of HT, ThT or any of the antioxidants affected cell viability during sulfide exposure, erythrocytes (diluted 1:1400 in incubation buffer) were pre-incubated with HT, ThT or an antioxidant for 1 h at 15°C in the microplates, after which H₂S gas was added, as described above, and the erythrocytes were incubated for an additional 2 h. Following the incubation, the cells were labeled with vital fluorescent dyes and assessed for viability, as described below. This experiment was independently replicated with erythrocytes from five worms, with erythrocyte samples from each worm being exposed to all concentrations of each compound under control conditions and at each level of H₂S (i.e. a fully balanced design).

To assess erythrocyte viability, a vital fluorescent dye mix was created in incubation buffer by adding calcein [loaded as calcein green acetoxy-methylester, a cell-permeant cytoplasmic stain that labels cells with intact plasma membranes; 10 mol l^–1 from a 2 mmol l^–1 stock solution in dimethyl sulfoxide (DMSO)], propidium iodide (PI, a cell-impermeant nucleic acid stain that labels cells with compromised plasma membrane integrity; 5 μgml^–1 from a 1 mg ml^–1 stock solution in H₂O) and Hoechst 33342(a cell-permeant nuclear stain that is used to identify cells; 2μgml^–1 from a 1 mg ml^–1 stock solution in H₂O). This dye mix was prepared daily from stock solutions stored at –20°C before use. Immediately after the erythrocytes were exposed to sulfide, as described above, 5 μl dye mix was added to each erythrocyte-containing well in each microplate and the dyes were allowed to load into the cells for 25 min in the dark at room temperature. After dye loading, 100μl incubation buffer was added to each well, which diluted any unloaded dyes to reduce background fluorescence. All dyes were from Invitrogen Corporation (Carlsbad, CA, USA).

For fluorescence detection of the fluorescent dyes, each microplate was placed on an epifluorescence microscope (Olympus IX-70, Olympus America Inc.,Center Valley, PA, USA) with a DAPI/FITC/TRITC triple bandpass polychroic emission filter set (Chroma Technology, Rockingham, VT, USA). The dyes were excited with light from a 100 W mercury bulb through appropriate excitation filters (484/15 nm for calcein, 555/15 nm for PI and 360/40 nm for Hoechst),and digital monochrome images were recorded with a cooled CCD camera (2000R,QImaging, Burnaby, BC, Canada). The images were later analyzed using ImageJ software (v. 1.37c; http://rsb.info.nih.gov/ij/). Cells were placed into three categories based on calcein and PI labeling: (1)cells that were labeled by calcein but not PI were considered viable; (2)cells that were labeled by PI with nuclear-specific staining were considered non-viable; and (3) cells that were labeled by PI with cytoplasmic staining were considered transitional and, although probably lethally injured, were not counted in the statistical analyses. The latter category was never present in erythrocytes from control conditions. Hoechst labeled both viable and non-viable erythrocytes and was used as a nuclear marker.

All data are presented as means ±1 standard error of the mean(s.e.m.), unless noted otherwise. The effects of HT and the antioxidants on dissolved sulfide concentration were analyzed by two-factor analysis of variance (ANOVA) followed by the Tukey HSD test (JMP 7.2, SAS Institute, Cary,NC USA). In the graphs, data for PI labeling represent both cytoplasmic and nuclear staining but statistical analysis of PI labeling was restricted to data for nuclear staining to be maximally conservative (see Discussion). The effects of H₂S on the fractions of cells labeled by calcein and PI were each analyzed as a split plot design, with antioxidant (including HT and ThT) as the whole plot factor, H₂S as the split plot factor and the interaction between animal and antioxidant as the error term for the whole plot factor. This design reduces the residual variation resulting from between-worm differences in cell viability. Because the fractions were not normally distributed (i.e. the majority of cells in any well were typically either viable or non-viable), the data were analyzed with a generalized linear mixed model (GLIMMIX, SAS software, SAS Institute). However, this analysis could not be performed for treatments that had zero variance (i.e. when all cells in a treatment were calcein-positive, calcein-negative, PI-positive or PI-negative) and, therefore, tests of statistical significance were limited to 0.10% H₂S for calcein and 0.32% H₂S for PI, which were selected because they are closest to the LD₅₀ for each dye (see Results).

The total dissolved sulfide concentration increased approximately proportionally to the H₂S gas fraction, as would be expected, with 0.10, 0.32, 1.0 and 3.2% H₂S producing 0.18, 0.32, 1.9 and 4.8 mmol l^–1 of sulfide, respectively(Fig. 1A, blue symbols). The addition of HT significantly reduced the dissolved sulfide concentration(whole model F_19,126=110, P<0.0001 for sulfide; P=1.0 for HT; P<0.0001 for sulfide×HT),with the addition of 50 mmol l^–1 of HT decreasing the sulfide concentration by 80% at 1.0% H₂S and 86% at 3.2% H₂S(Fig. 1A, red symbols). By contrast, adding the highest concentration of the antioxidants did not significantly affect the dissolved sulfide concentration(Fig. 1B–F), with the exception of 3.0 mmol l^–1 Tempol at 1.0% H₂S,which reduced the sulfide concentration by 50%. The general effect of HT was concentration-dependent and showed a similar trend at each H₂S percentage (Fig. 1G),consistent with simple kinetics and the absence of saturation.

G. dibranchiata erythrocytes exposed to control conditions (air)showed cytoplasmic calcein labeling but no PI labeling(Fig. 2A), consistent with viable cells that have an intact plasma membrane. Exposure to 0.10%–3.2%H₂S gas (in air) caused a dose-dependent decrease in labeling by calcein and increase in nuclear labeling by PI(Fig. 2B–E). This is consistent with a progressive loss of plasma membrane integrity leading to a net outward diffusion of calcein and an inward diffusion of PI (and its subsequent labeling of nucleic acids). Quantification of the number of cells labeled by each dye showed that H₂S exposure produced a significant decrease in the fraction of cells labeled with calcein green(Fig. 2F) (green circles; F_4,8=21.1, P=0.0003, N=3) with the calcein-positive fraction decreasing from 0.92±0.03 in the controls to 0.01±0.01 at 3.2% H₂S (P=0.0005). Similarly,H₂S exposure produced a significant increase in the fraction of cells labeled by PI (Fig. 2F,red circles) (F_4,8=83.5, P<0.0001, N=3) with the total PI-positive fraction increasing from 0.04±0.03 in the controls to 0.99±0.01 at 3.2% H₂S(P<0.0001), and with the subset of nuclear PI-positive cells increasing from 0.04±0.03 in the controls to 0.55±0.20 at 3.2%H₂S (Fig. 2F, red triangles) (F_4,8=9.93, P<0.003, N=3). At the highest sulfide concentration, PI labeling was almost exclusively nuclear whereas cytoplasmic PI labeling (whether alone or with nuclear labeling) was more common at intermediate sulfide concentrations (data not shown). Based on the calcein, nuclear PI, and total PI labeling, the LD₅₀ for H₂S appeared closest to 0.32%.

Addition of HT and ThT to the incubation buffer did not affect cell viability under control conditions (i.e. the absence of H₂S),indicating that neither of these compounds was toxic on its own. Addition of HT increased the viability of cells exposed to H₂S, as evident from a general increase in calcein labeling and decrease in PI labeling(Fig. 3, HT panels A–D). As the concentration of HT increased, the fraction of cells labeled with calcein significantly increased (Fig. 3, HT panel E) (F_3,11=38.4, P<0.0001, N=4) whereas the fraction of cells showing nuclear PI labeling significantly decreased(Fig. 3, HT panel F)(F_3,11=3.87, P<0.041, N=4). For example, adding 50 mmol l^–1 of HT (the highest concentration tested) increased the fraction of viable cells from 0.01 to 0.84 at 0.32%H₂S, from 0.00 to 0.79 at 1.0% H₂S, and from 0.02 to 0.21 at 3.2% H₂S. To determine the statistical significance of adding HT, we performed a GLIMMIX analysis at 0.10% H₂S for calcein labeling and 0.32% H₂S for nuclear PI labeling(Table 1). These two H₂S concentrations were selected because higher and lower concentrations tended to produce zero variance (i.e. all cells were positive or negative for one of the dyes at a particular H₂S concentration in at least one replicate) and, therefore, did not allow GLIMMIX analysis. At 0.10% H₂S, calcein labeling was significantly increased at all concentrations of HT (0.5, 5.0 and 50 mmol l^–1)(Table 1) and at 0.32%H₂S, nuclear PI labeling was significantly decreased at 5.0 and 50 mmol l^–1 HT but not 0.50 mmol l^–1 HT(Table 1).

In contrast to the effect of HT, adding increasing concentrations of ThT generally decreased labeling by calcein and increased labeling by PI in cells exposed to H₂S (Fig. 3, ThT panels A–D). Adding ThT at 0.50 and 5.0 mmol l^–1 significantly decreased the viability of cells exposed to H₂S, based on both calcein labeling(Fig. 3, ThT panel E; Table 1)(F_2,5=49.9, P<0.0005, N=3) and PI labeling (Fig. 3, ThT panel F; Table 1)(F_2,5=16.0, P<0.006, N=3). For example, the addition of 5 mmol l^–1 of ThT at 0.10%H₂S decreased calcein labeling from 0.95 to 0.40 whereas the addition of 5 mmol l^–1 of ThT at 0.32% H₂S increased nuclear PI labeling from 0.12 to 0.39.

None of the antioxidants affected cell viability under control conditions,indicating that these were not toxic on their own. Generally, adding GEE increased labeling by calcein and decreased labeling by PI in cells exposed to H₂S (Fig. 4 GEE,panels A–D). Adding GEE at 0.030 and 0.30 mmol l^–1 but not 3.0 mmol l^–1, significantly increased the viability of cells exposed to H₂S, based on both calcein labeling(Fig. 4, GEE panel E; Table 1)(F_3,15=10.6, P<0.0005, N=4) and PI labeling (Fig. 4, GEE panel F; Table 1)(F_3,15=6.88, P<0.004, N=4). Similarly, the addition of increasingly higher concentrations of NAC increased labeling by calcein and decreased labeling by PI in cells exposed to H₂S (Fig. 4, NAC panels A–D). All concentrations of NAC (0.030, 0.30, 1.0 mmol l^–1) significantly increased cell viability, based on both calcein labeling (Fig. 4, NAC panel E; Table 1)(F_3,13=17.4, P<0.0001, N=4) and PI labeling (Fig. 4, NAC panel F; Table 1)(F_3,13=43.6, P<0.0001, N=4).

The addition of higher concentrations of ASC, Tempol or Trolox either had no effect or decreased the viability of G. dibranchiata erythrocytes exposed to H₂S. Adding increasingly higher concentrations of ASC generally decreased labeling by calcein and increased labeling by PI in cells exposed to H₂S (Fig. 4, ASC panels A–D). Adding ASC at 0.010 and 0.10 mmol l^–1 significantly decreased the viability of cells exposed to H₂S, based on both calcein labeling(Fig. 4, ASC panel E; Table 1)(F_3,10=114, P<0.0001, N=4) and PI labeling (Fig. 4, ASC panel F; Table 1)(F_3,10=49.1, P<0.0001, N=4). Similarly, the addition of higher concentrations of Tempol generally decreased labeling by calcein and increased labeling by PI in cells exposed to H₂S (Fig. 4, Tempol panels A–D). Adding Tempol at 0.30 and 3.0 mmol l^–1significantly decreased the viability of cells exposed to H₂S,based on both calcein labeling (Fig. 4, Tempol panel E; Table 1) (F_3,15=112, P<0.0001, N=4) and PI labeling (Fig. 4, Tempol panel F; Table 1) F_3,15=15.1, P<0.0001, N=4). Addition of Trolox did not substantially change calcein labeling or PI labeling in cells exposed to H₂S(Fig. 4, Trolox panels A–D) and, therefore, did not affect cell viability(Fig. 4, Trolox panel E–F; Table 1).

In previous studies, in vitro exposure of G. dibranchiataerythrocytes to sulfide for 1 h caused dose-dependent mitochondrial depolarization with an EC₅₀ of approximately 0.5–0.8 mmol l^–1, increased cellular oxidative stress and increased superoxide production (Julian et al.,2005); in vivo exposure of G. dibranchiata worms to sulfide for 24 h caused a dose-dependent increase in PI labeling of erythrocytes and a decrease in cell proliferation at sulfide concentrations above 0.5 mmol l^–1 (Hance et al., 2008); and sulfide exposure caused oxidative damage to RNA and DNA in G. dibranchiata erythrocytes exposed to sulfide in vitro and in erythrocytes and body wall of G. dibranchiataexposed to sulfide in vivo(Joyner-Matos et al., 2008). However, following in vivo exposure, G. dibranchiata showed no change in gross morphology or behavior at sulfide concentrations up to 1 mmol l^–1, and short-term survival of the animals was not affected at sulfide concentrations as high as 2.4 mmol l^–1(Hance et al., 2008). Consequently, although G. dibranchiata can clearly be classified as`sulfide-tolerant' based on survival during sulfide exposure, the erythrocytes show evidence of cytotoxicity and oxidative damage both in vivo and in vitro. These findings are consistent with sulfide exposure causing increased free radical production and oxidative stress(Abele-Oeschger and Oeschger,1995; Attene-Ramos et al.,2007; Chen and Morris,1972; Eghbal et al.,2004; Joyner-Matos et al.,2006; Tapley,1993; Tapley et al.,1999; Truong et al.,2006) and leads to the hypothesis that appropriate antioxidants could increase the viability of cells exposed to sulfide.

To assess erythrocyte viability in the present study, we determined whether the cells became labeled with the fluorescent dyes calcein and PI after H₂S exposure. Calcein (or other similar cell-permeant fluorophores)and PI (or other similar cell-impermeant fluorophores) have long been used together as `live/dead stains' in mammalian cells in vitro(Dive et al., 1990) but these have rarely been used together in marine invertebrates. However, hemocyte viability has been assessed in bivalves and crustaceans using calcein alone(Cardenas et al., 2004; Marin et al., 2004) and PI alone (Hegaret et al., 2003; Lee et al., 2001; Xue et al., 2001), and in G. dibranchiata erythrocytes in vivo using PI alone(Hance et al., 2008). We found that under control conditions in vitro, 100% of G. dibranchiata erythrocytes were labeled by calcein but were not labeled by PI whereas at 3% H₂S (approximately 5 mmol l^–1dissolved sulfide), 99% of the cells were labeled by PI but were not labeled by calcein. Cytoplasmic PI labeling, whether alone or with nuclear labeling,was more common at intermediate sulfide concentrations whereas PI labeling of cells exposed to the highest H₂S concentration was almost exclusively nuclear. This is consistent with previous studies indicating that non-nuclear PI labeling represents an intermediate membrane permeability state(Dive et al., 1990). Nonetheless, to be maximally conservative, statistical analyses of PI labeling were performed with data for nuclear PI labeling, for which the association with loss of cell viability is least ambiguous. These results are consistent with H₂S exposure compromising plasma membrane integrity, whether directly or indirectly. A half-maximal effect of H₂S on cell viability (LD₅₀) was seen at 0.32% H₂S, which produces approximately 0.3 mmol l^–1 dissolved sulfide. Interestingly,this is similar to the EC₅₀ of 0.5–0.8 mmol l^–1 sulfide for erythrocyte mitochondrial depolarization in vitro (Julian et al.,2005) and the erythrocyte cell death threshold of 0.5 mmol l^–1 sulfide in vivo(Hance et al., 2008). This suggests a common mechanism of toxicity at moderate sulfide concentrations,which has yet to be determined. Between-worm variability was evident in the effect of sulfide on cell viability. For the statistical analyses, this residual variation was reduced by the split plot design, in which samples of cells from each animal were exposed to every H₂S concentration (and at every concentration of HT, ThT and the antioxidants). The source of this inter-individual variation is unknown but such differences in the response of cells to sulfide in vitro could produce variation in whole-animal sulfide tolerance, which would be likely to affect features of populations such as the distribution within a sulfidic habitat.

HT has been hypothesized to protect deep-sea animals from sulfide toxicity by scavenging free radicals generated from sulfide oxidation and/or by scavenging sulfide (Alberic,1986; Alberic and Boulegue,1990; Horak et al.,2003; Pruski et al.,1997; Pruski et al.,2000b; Rosenberg et al.,2006; Yancey,2005; Yancey et al.,2002b; Yin et al.,2000) but whether HT reduces sulfide toxicity and, if so, by what mechanism, had never been tested. We found that incubation buffer containing HT had up to 86% less dissolved sulfide after 2 h exposure to H₂S than incubation buffer without HT, and that HT increased the viability of G. dibranchiata erythrocytes exposed to H₂S for 2 h. For example at 1% H₂S in the absence of HT, no cells were calcein-positive, whereas 83% of cells were calcein-positive at 1%H₂S with 50 mmol l^–1 HT. The effect on PI-labeling was similar but less dramatic. Overall, 50 mmol l^–1 HT shifted the LD₅₀ from 0.32% H₂S to 1–3%H₂S, with lower HT concentrations having a smaller but still significant effect. Considered alone, the cell viability data do not distinguish whether the mechanism of protection is via scavenging free radicals, scavenging sulfide or both. However, the positive effect of HT on cell viability was roughly correlated with its negative effect on dissolved sulfide concentration. This suggests, but does not prove, that sulfide scavenging is the most important factor in the ability of HT to reduce sulfide toxicity.

Because the ability of HT to reduce the dissolved sulfide concentration was observed in cell-free incubation buffer, it was independent of enzymatic activity. It would be valuable to test whether this effect is stronger in tissues and, in particular, whether tissues of sulfide-adapted animals that accumulate HT have specific enzymatic pathways that promote sulfide scavenging by HT. Because we did not measure HT or ThT concentrations, we also do not know the kinetics, equilibrium constant or other characteristics of the reaction between sulfide and HT but, clearly, these would be valuable to obtain, especially in tissues. Interestingly, although HT is assumed to be cell-permeant in its zwitterionic form(Alvarez and Storey, 1983), it is typically concentrated primarily within cells, with lower concentrations in hemolymph (Fiess et al., 2002; Pruski et al., 2000a; Yin et al., 2000), which indicates that it is not distributed strictly by diffusion. It is unknown how HT was ultimately distributed across the erythrocyte plasma membrane in our experiments but it is reasonable to assume that HT reduced the intracellular sulfide concentration as well as the extracellular sulfide concentration.

As the reaction product of sulfide and HT, ThT has no further ability to bind sulfide. However, if ThT is an antioxidant(Cavallini and Tentori, 1960)it may still reduce sulfide toxicity by binding free radicals from sulfide oxidation. Therefore, we anticipated that ThT would at worst have no effect on cell viability during H₂S exposure and at best might have a moderate beneficial effect. However, while ThT did not affect the viability of cells under control conditions (and was therefore not toxic itself), it decreased cell viability during H₂S exposure by as much as 83% (at 0.5 mmol l^–1 ThT and 0.10% H₂S). The mechanism by which ThT increased H₂S toxicity is unknown but any antioxidant capacity it might have had was not sufficient to protect the cells. Furthermore, this raises the question of whether, in cells incubated with HT during sulfide exposure, gradual accumulation of ThT begins to reduce cell viability. This exemplifies the importance of recycling of ThT to HT, whether by sulfide-oxidizing endosymbionts or by specific enzymes(Brand et al., 2007; Rosenberg et al., 2006).

To further explore whether free radical scavenging alone could decrease the toxicity of sulfide in G. dibranchiata erythrocytes, we tested whether the antioxidants GEE, NAC, ASC, Tempol and Trolox could reduce the dissolved sulfide concentration in cell-free buffer exposed to H₂S and increase the viability of cells exposed to H₂S. Other than ASC,these antioxidants are all cell-permeant. Unlike HT, none of the antioxidants significantly decreased the dissolved sulfide concentration in incubation buffer exposed to H₂S but all concentrations of GEE and NAC(0.03–3 and 0.03–1 mmol l^–1, respectively)increased the fraction of viable cells. This suggests that the beneficial action of GEE and NAC was at least partially via free radical scavenging, although an alternative explanation is that GEE and NAC reduced free sulfide primarily intracellularly, which would not have been detected in our experiments. GEE is converted within cells to the antioxidant reduced glutathione (Halliwell and Gutteridge,2007), which, upon becoming oxidized, can react with sulfide,forming glutathione persulfide (Rohwerder and Sand, 2003; Smith and Gosselin, 1966). NAC reacts with reactive oxygen species to produce cysteine (Aruoma et al.,1988) and stimulates intracellular glutathione synthesis(Moldeus et al., 1986) but it also contains a sulfhydryl group. The remaining antioxidants, all of which do not contain a sulfhydryl group, either reduced the viability of cells exposed to H₂S or had no effect. ASC readily oxidizes to dehydroascorbic acid, reducing the oxidative loss of sulfide in vitro(Glaister et al., 1984). All concentrations of ASC (0.001–0.1 mmol l^–1) decreased cell viability. The superoxide dismutase-mimetic Tempol is a catalytic scavenger of superoxide anions in vitro. Tempol at lower concentrations (0.03–0.3 mmol l^–1) had no effect on cell viability but 3 mmol l^–1 Tempol decreased cell viability in the presence of H₂S (e.g. by as much as 77% at 0.10%H₂S), consistent with a report that Tempol can be deleterious at concentrations higher than 0.05 mmol l^–1(Lewinska et al., 2008). Trolox, which inhibits membrane lipid peroxidation and protein carbonylation(Halliwell and Gutteridge,2007), had no effect on cell viability at any concentration(0.001–0.1 mmol l^–1) but it is important to note that we used comparatively low concentrations so it remains possible that higher concentrations would have been more effective. Interestingly, the persulfide dithiothreitol (DTT) is an antioxidant that contains two sulfhydryl groups and has been reported to reduce the lethality of H₂S exposure in rats,presumably by reducing the free sulfide concentration(Reiffenstein et al., 1992). We tested the ability of 0.05–5.0 mmol l^–1 DTT to increase cell viability during H₂S exposure and observed a similar effect to that of GEE and NAC; however, the highest DTT concentration interfered with the loading, retention or fluorescence of the dyes through an unknown mechanism and therefore the DTT data were excluded.

The ability of HT to reduce sulfide toxicity was much greater than that of GEE and NAC but the maximum concentration of HT we used was 16-times and 50-times higher than that of GEE and NAC, respectively. HT is highly biocompatible, and the HT concentrations we used were within the range of tissue concentrations reported for many animals from hydrothermal vents,although it is somewhat difficult to make comparisons with the literature because the methods used to measure and express the HT concentration have varied. For example, when measured as HT per tissue wet mass, reported HT concentrations are 63–152 mmol kg^–1 in several tubeworms (Yin et al., 2000)and 59–63 mmol kg^–1 in two gastropods(Rosenberg et al., 2006). When measured as HT per tissue dry mass, reported concentrations are 7.2–112 mmol kg^–1 in a large variety of bivalves and vestimentiferans(Fiess et al., 2002; Pruski et al., 2000a). Based on these values, Yancey estimated cell solute HT concentrations of 20–60 mmol l^–1 HT for Calyptogena sp. and 320–360 mmol l^–1 HT for Riftia pachyptila(Yancey, 2005).

HT, and to a lesser extent the sulfhydryl-containing antioxidants GEE and NAC, increase the viability of G. dibranchiata erythrocytes exposed to H₂S in vitro. The data support the hypothesis that HT can protect tissues of deep-sea, sulfide-adapted animals from sulfide cytotoxicity and suggest that sulfide scavenging rather than free radical scavenging is the most important mechanism of protection, although this does not rule out free radical scavenging as an additional protective mechanism. There are no reports of shallow water invertebrates with high HT concentrations (Fiess et al.,2002; Yancey,2005; Yancey et al.,2002a) but to our knowledge animals in shallow sulfidic habitats have not been specifically investigated. Consequently, it is still unknown whether HT or similar compounds may reduce sulfide toxicity in G. dibranchiata or other invertebrates living in characteristically sulfidic habitats such as mudflats and mangroves. Finally, it is evident that H₂S, from both exogenous and endogenous sources, has a variety of physiological actions (Szabó,2007) and, therefore, it would be valuable to determine the extent to which HT or compounds with sulfhydryl groups may modulate the activity of H₂S in vertebrate tissues.

LIST OF ABBREVIATIONS

 
• ASC

  l-ascorbic acid

 
• DTT

  dithiothreitol

 
• GEE

  glutathione ethyl ester

 
• HT

  hypotaurine

 
• NAC

  N-acetylcysteine

 
• PI

  propidium iodide

 
• Tempol

  4-hydroxy-2,2,6,6-tetramethylpiperidine-1-oxyl

 
• ThT

  thiotaurine

 
• Trolox

  6-hydoxy-2,5,7,8-tetramethylchroman-2-carboxylic acid

J.A.O. and J.M.O. contributed equally to this manuscript. This research was supported by NSF IBN-0422139 to D.J. The authors thank Meghan Brennan,University of Florida Department of Statistics, for providing consultation on the statistical analyses. All experiments complied with the requirements of the University of Florida Institutional Animal Care and Use Committee.