Sulfide Acquisition By The Vent Worm Riftia Pachyptila Appears To Be Via Uptake Of Hs⁻, Rather Than H₂S

The hydrothermal vent tubeworm Riftia pachyptila was first found to be symbiotic with intracellular carbon-fixing sulfide-oxidizing bacteria in 1981 (Cavanaugh et al. 1981; Felbeck et al. 1981). Because R. pachyptila relies upon these internal bacterial symbionts for its nutrition, it must supply them with carbon dioxide, oxygen, hydrogen sulfide and other nutrients (Fisher, 1990; Childress and Fisher, 1992). These compounds are taken up from the environment across the plume and transported, via a well-developed vascular system, to the bacteria, found in a highly vascularized organ known as the trophosome (Jones, 1981; Arp et al. 1985; Childress and Fisher, 1992). This organ is located within the trunk of the worm, surrounded by non-circulating coelomic fluid, which is apparently in equilibrium with the circulating vascular blood for smaller molecules such as CO₂, H⁺ and H₂S (Childress et al. 1984, 1991).

These worms have two extracellular hemoglobins in the vascular blood and another in the coelomic fluid that bind and transport both oxygen and hydrogen sulfide to the symbiont (Arp et al. 1985, 1987; Childress et al. 1991; Zal et al. 1996a,b). Vascular blood has a higher capacity for sulfide and contains more total sulfide (ΣH₂S) than coelomic fluid because the larger of the two vascular hemoglobins has a higher capacity (approximately threefold higher) for sulfide binding and overall hemoglobin concentration in the vascular blood is higher (Arp et al. 1987; Childress et al. 1991). R. pachyptila body fluids can reach extremely high concentrations of ΣH₂S (including H₂S, HS⁻ and S²⁻), up to 6 mmol l⁻¹ in the coelomic fluid and 12 mmol l⁻¹ in the vascular blood (Childress et al. 1991). Other than the binding of sulfide by the hemoglobins, however, the mechanism of sulfide uptake has not been studied.

Goffredi et al. (1997) proposed that the mechanism for inorganic carbon acquisition in these worms is diffusion of the undissociated CO₂ species, which is supported by effective control of body fluid pH by proton-equivalent export, rather than mediated uptake of HCO₃⁻. Both carbon dioxide and hydrogen sulfide demonstrate strong pH-dependent dissociation. The pK values, or dissociation constants (i.e. the pH values at which the ratios CO₂:HCO₃⁻ and H₂S:HS⁻ are 1:1), for CO₂ and H₂S are 6.1 and 6.6, respectively (Dickson and Millero, 1987; Millero et al. 1988), at 10 °C and 101.3 kPa. Thus, H₂S, like CO₂, is the dominant chemical species in situ owing to the acidic pH (near 6.0) around the worms. H₂S diffusion into the worms would be a function of the external and internal (intra- and extracellular) concentrations of free H₂S, which are functions of ΣH₂S, pH and sulfide binding by the hemoglobins in the blood. Hypothetically, if sulfide were acquired via H₂S diffusion, sulfide would be concentrated in the blood, as is the case for inorganic carbon because, at the physiological pH of 7.1–7.5 of R. pachyptila, H₂S dissociates into HS⁻ and H⁺ and the protons would normally be eliminated (Childress et al. 1984, 1991; Goffredi et al. 1997). One would predict in this case that, all else being unchanged, a lower external pH would increase the rate of sulfide uptake while a lower internal (extracellular) pH would decrease uptake.

In contrast, if the mechanism for sulfide acquisition in R. pachyptila were HS⁻ uptake, the pH-dependence of this mechanism would be expected to differ from that of CO₂ and H₂S uptake. HS⁻ uptake would probably be mediated by the negative charge of the ion. If it were to occur via facilitated diffusion, uptake of HS⁻ might increase with decreases in extracellular pH, because of effects on the equilibrium in the body fluids. It is possible, however, that the uptake of HS⁻ may not be affected much by changes in internal hydrogen ion concentration.

The purpose of the present study was to determine the mechanism used by R. pachyptila to acquire sulfide from the environment. This involved the measurement of internal sulfide concentrations from freshly captured worms and the execution of live animal experiments in pressure systems on board ship. To differentiate between H₂S diffusion and mediated uptake of HS⁻, we measured coelomic fluid and vascular blood ΣH₂S of worms in two types of experiments: in one, we varied external H₂S and HS⁻ levels around the worms; in the other, we forced a decrease in the extracellular pH of the worms by exposing them either to hypoxic water or to a non-specific inhibitor of H⁺-ATPases, N-ethylmaleimide.

Riftia pachyptila Jones were collected at a mean depth of 2600 m by submersible (D.S.R.V. Alvin and Nautile) during research expeditions to 9°N (9°50′N, 104°18′W) and 13°N (12°48′N, 103°57′W) along the East Pacific Rise in 1994 and 1996. In 1994, hot venting water and warm water samples around the tubeworms were collected using titanium samplers from the Alvin. Animals were brought to the surface in a temperature-insulated container and transferred to cold sea water (5 °C) in a refrigerated van on board ship. Worms were then sorted to be used either for experiments on living animals or for immediate measurements of physiological parameters. Live animal experiments were initiated within 2 h of surfacing.

All experiments were conducted inside a refrigerated van.

Sea water was chilled by moving the water through polypropylene tubing past a refrigeration unit, after which it was pumped by high-pressure, Teflon diaphragm, metering pumps through stainless-steel vessels at flow rates ranging from 4 to 12 l h⁻¹, at a pressure of approximately 21.5 MPa. Pressure gauges and sample ports were placed in-line immediately after flow through the vessels to allow monitoring of pressure and water conditions. All worms were kept in these pressurized flowing-water aquaria, in which we were able to re-create many aspects of the vent environment, such as temperature, pressure and a variety of chemical conditions, including pH, ΣCO₂, ΣH₂S, O₂ and N₂ concentrations (Quetin and Childress, 1980; Goffredi et al. 1997). Water ΣCO₂, O₂ and N₂ levels and were varied by bubbling CO₂, O₂ and N₂ gas directly into a gas equilibration column, which supplied water to the high pressure pumps (see diagram in Kochevar et al. 1992). Sulfide concentrations were controlled by continuously pumping anaerobic solutions of sodium sulfide (30–50 mmol l⁻¹) into the gas equilibration column at rates dependent upon the desired final concentrations of sulfide. At the end of each experiment, the animals were quickly removed from the pressure vessel and dissected.

In one type of experiment (‘sulfide series’), worms were placed in experimental vessels immediately upon collection from the sea floor. Thirty worms were kept at 8 °C and exposed to 4.0±0.5 mmol l⁻¹ ΣCO₂ (mean ± S.E.M. for all values given) and 191±18 μmol l⁻¹ O₂ for 17–20.5 h. To stabilize the pH, 10 mmol l⁻¹ Mops or Mes buffer was added to the sulfide solutions. In some cases, depending on the desired final pH value, it was necessary to alter the pH of the buffered solutions by titration with hydrochloric acid. In this particular experiment, we achieved external [H₂S] values between 0 and 362 μmol l⁻¹ and [HS⁻] values between 0 and 265 μmol l⁻¹ by controlling the external pH between 5.59 and 7.21 and ΣH₂S values between 63 and 511 μmol l⁻¹.

In another experiment (‘hypoxia’ experiments), we exposed twelve worms to hypoxic conditions (at 15 °C) with external oxygen levels no greater than 42 μmol l⁻¹ O₂ for 13 h, while control worms were kept at oxygen levels of 316±23 μmol l⁻¹. In situ concentrations of O₂ around the worms fluctuate due to the mixing of vent waters (0 μmol l⁻¹) with ambient water (110 μmol l⁻¹). Data on the metabolism of the worms and the distribution of O₂ around them, however, show that they take up O₂ primarily from concentrations approaching 100 μmol l⁻¹ (Johnson et al. 1988; Childress et al. 1991). In our experience, O₂ concentrations between 100 and 400 μmol l⁻¹ do not appear to affect the symbioses (P. Girguis, personal communication). All worms in these experiments were kept in the aquaria described above and exposed to typical vent water concentrations of ΣCO₂ (4.4±0.2 mmol l⁻¹), ΣH₂S (156±21 μmol l⁻¹) and pH (6.23±0.05).

In additional experiments (‘inhibitor’ experiments), we initially kept worms in flowing-water maintenance aquaria, also at 8 °C and 21.5 MPa, supplied with surface sea water (ΣCO₂ levels of 2.1 mmol l⁻¹, pH 8.2, and no sulfide). This experiment, in which we needed a uniform starting point for all of the worms, was directed at measuring the rates of uptake of CO₂ and sulfide over time; thus, we needed these worms to have low internal levels of ΣCO₂ and ΣH₂S, which resulted from maintenance in surface sea water. In order to observe uptake rates over time, the animals were then transferred to experimental vessels (for 0–20 h), and water conditions (ΣCO₂ 4.8±0.1 mmol l⁻¹, O₂ 230±9 μmol l⁻¹, ΣH₂S 311±39 μmol l⁻¹ and pH 5.96±0.05) were controlled in the same manner as described for the sulfide series and hypoxia experiments. 15 worms were used as controls and 9 worms were exposed to N-ethylmaleimide (NEM), a non-specific inhibitor of H⁺-ATPases, at concentrations between 1.1 and 1.6 mmol l⁻¹ for 1–2 h (Marver, 1984; Stone et al. 1984).

It is important to note that during the 1–12 h of these inhibitor experiments, it is likely that the worms were not in autotrophic balance. It has been suggested that it takes at least 14 h in the presence of sulfide for these worms to become autotrophic (net uptake of CO₂; Childress et al. 1991). Another study has shown that, at 12 h, and under conditions similar to ours, there is some assimilation of both ¹³CO₂ and ¹⁵NO₃⁻ by R. pachyptila, indicating that the symbionts are functioning (Lee and Childress, 1994). Regardless, time points before 12 h are meaningful in terms of the functioning of the animal, and for our studies it was more important to isolate the physiology of the worms from that of the symbionts.

All worm dissection techniques were similar to those described in Childress et al. (1991) and Goffredi et al. (1997). ΣCO₂ and ΣH₂S of coelomic fluid, vascular blood and water samples were measured on 0.5 ml samples using a Hewlett Packard 5880A series gas chromatograph (Childress et al. 1984). ‘ΣH₂S’, as measured by the gas chromatograph, represents the sum of H₂S, HS⁻, S²⁻ concentrations and the amount of sulfide bound to the hemoglobins. ‘ΣCO₂’, as measured by the gas chromatograph, represents the sum of CO₂, HCO₃⁻ and CO₃²⁻ concentrations. Body fluid and water pH were measured using a thermostatted Radiometer BMS-2 blood pH analyzer equipped with a G299A capillary pH electrode and connected to a PHM73 pH meter. Additional water pH measurements were made using a double-junction combination electrode (Broadley-James) connected to a PHM93 pH meter (Radiometer).

The Kendall rank correlation was used to test for correlations. The Mann–Whitney U-test was used to test for differences in distribution between data sets. Simple regressions were used to show linear relationships, and multiple regressions were used to compare influences of various parameters. The analysis of covariance (ANCOVA) was used to compare slopes and magnitudes of different data sets.

Both freshly collected and experimental worms showed an increase in internal sulfide levels as the surrounding water pH increased. For freshly collected worms, the ΣH₂S values in both coelomic fluid and vascular blood increased as the surrounding water pH increased (P=0.0025 and P=0.0151, respectively, Kendall rank correlation). For ‘sulfide series’ worms, the coelomic fluid and vascular blood ΣH₂S levels also increased as we increased the surrounding water pH (P=0.0007 and P=0.0013, respectively, Kendall rank correlation). In addition, the internal ΣH₂S levels of these worms were positively correlated with the extracellular pH (y=3.451x−23.676, P=0.0006, and y=6.661x−43.014, P<0.0001 for the regressions of coelomic fluid and vascular blood of ‘sulfide series’ worms, respectively, where y is ΣH₂S and x is extracellular pH).

Worms collected from the sea floor and placed immediately into ‘sulfide series’ experiments had initial internal ΣH₂S levels of 0.18±0.12 mmol l⁻¹ and 0.43±0.17 mmol l⁻¹ for coelomic fluid and vascular blood, respectively. In order to determine which external species of sulfide influenced internal ΣH₂S levels, we controlled the free H₂S and HS⁻ concentrations in the surrounding water for 17–20.5 h. It should be noted that there was no correlation between water H₂S and HS⁻ levels during these experiments (P=0.55, Kendall rank correlation). In order to determine whether this 3.5 h variation among experiments played an important role in the values measured, we plotted internal sulfide concentration versus the incubation time of these experiments. Both graphs (data not shown) showed no dependence of internal sulfide levels on incubation time over the limited range of times used in our experiments (P=0.95 and P=0.74 for the regressions of vascular blood and coelomic fluid, respectively). This time, however, far exceeds the time necessary to reach sulfide equilibrium with the surrounding water, as considered below.

Fig. 1 shows that both coelomic fluid and vascular blood ΣH₂S levels correlated well with external HS⁻ (P<0.0001), but not with H₂S (P=1). A multiple regression was used to determine whether H₂S or HS⁻ level in the external medium had a greater influence on internal ΣH S values. It was apparent from this test that HS⁻ plays a greater role (P=0.0001) in predicting coelomic fluid ΣH₂S than does H₂S (P=0.0901) and also in predicting vascular blood ΣH₂S values (P=0.001 for HS⁻ and P=0.81 and H₂S). Both the coelomic fluid and vascular blood pH increased as water HS⁻ increased (y=1.875x+7.287, P=0.0058, and y=2.411x+7.182; P=0.0018, respectively, for the regression equations, where y is pH and x is water [HS⁻]). A multiple regression analysis was used to determine whether H₂S or HS⁻ level in the external medium had a greater influence on extracellular pH. Again, HS⁻ plays a greater role in predicting coelomic fluid and vascular blood pH (P=0.0124 and P=0.0052, respectively) than does H₂S (P=0.39 and P=0.21, respectively).

To test the effect of body fluid pH upon sulfide uptake, we conducted two types of whole-animal experiments in which we forced a decrease in the extracellular pH of the worms. In the first experiment, twelve worms were exposed to hypoxic conditions with external oxygen levels no greater than 42 μmol l⁻¹ O₂. Exposure to these low oxygen levels resulted in significantly decreased extracellular pH (Table 1), probably because of the build up of end-products of anaerobic metabolism. However, this decrease in extracellular pH failed to significantly affect internal ΣH₂S levels in these worms (Table 1). The situation for internal ΣCO₂ was reversed in that the decrease in extracellular pH caused a significant decrease in body fluid ΣCO₂ levels (Table 1). This was expected for ΣCO₂ because inorganic carbon is concentrated into the worms as a result of the pH difference maintained between the internal and external fluids (Goffredi et al. 1997). More importantly, the mechanism for sulfide acquisition appears to have a different pH-dependence from that for inorganic carbon acquisition.

In the second experiment, in which 9 worms were exposed to N-ethylmaleimide and 15 were used as controls, we followed the accumulation of sulfide and inorganic carbon in the worms over time. After 2–4 days in water with a low and no sulfide, coelomic fluid ΣCO₂ levels were 3.8±0.7 mmol l⁻¹ and ΣH₂S levels were 0.002±0.001 mmol l⁻¹ (worms at time zero). When subsequently exposed to flowing water at typical vent conditions (4.8±0.1 mmol l⁻¹ ΣCO₂ and 0.31±0.04 mmol l⁻¹ ΣH₂S), these worms demonstrated increases in total extracellular fluid inorganic carbon and sulfide. Between 0 and 7 h, the rate of increase in coelomic fluid ΣCO₂ in control worms was 3.22 mmol l⁻¹ h⁻¹, with the rate leveling off after 7 h at a mean ΣCO₂ concentration of 27.8 mmol l⁻¹ (Fig. 2A) (see also Goffredi et al. 1997). NEM-treated worms, however, did not accumulate inorganic carbon in the blood as rapidly (2.2 mmol l⁻¹ h⁻¹) and ΣCO₂ did not plateau at the same level (17.1 mmol l⁻¹; Fig. 2A). In addition, the control worms demonstrated a sulfide uptake rate of 0.97 mmol l⁻¹ h⁻¹ within the first hour (Fig. 2B). There appeared to be no further increase in the internal ΣH₂S of control worms (mean 0.81±0.14 mmol l⁻¹) after 1 h. There was, however, a slight increase (from 1.59 to 2.55 mmol l⁻¹ at a rate of 57 μmol l⁻¹ h⁻¹) in the internal ΣH₂S of NEM-treated worms over time (P=0.0211; Fig. 2B). The mean external sulfide concentrations for the two groups were not significantly different (Mann-Whitney test, 0.27±0.05 mmol l⁻¹ for control and 0.37±0.07 mmol l⁻¹ for NEM-treated worms); however, internal ΣH₂S and external ΣH₂S for both groups of worms are not correlated over the limited range in this data set (P=0.28, Kendall rank correlation). Thus, the difference in external ΣH₂S levels cannot explain the difference in internal ΣH₂S levels measured between the two groups. After 4 h, the increase in coelomic fluid ΣH₂S in the NEM-treated worms was not significantly different from that in the control worms (ANCOVA, P=0.1650); however, they were significantly different in overall magnitude (ANCOVA, P=0.0013).

It has been proposed that regulation of extracellular pH in R. pachyptila occurs primarily through proton-equivalent ion transport via an ATP-requiring process, specifically via H⁺-ATPases (Goffredi et al. 1997). The apparent inhibition of proton-equivalent ion transport by NEM resulted in significant decreases of approximately 0.4 pH units in both coelomic fluid and vascular blood pH (Table 2). This decrease in extracellular pH, however, did not result in a decrease in internal ΣH₂S levels, as expected in the case of H₂S diffusion, but rather a significant increase in internal sulfide levels (Table 2). Again, this result is contrary to that seen for internal ΣCO₂, in which a lower extracellular pH caused a reduction in the diffusion gradient for CO₂, resulting in significant decreases in internal ΣCO₂ (Table 2). These contrasting results support different

modes of acquisition for inorganic carbon and sulfide.

On the basis of our studies, we propose a model for sulfide uptake in R. pachyptila in which sulfide acquisition from the environment is primarily via HS⁻ uptake facilitated by transporter or channel proteins. Once across the outer epithelium of the plume, HS⁻ must enter the vascular blood compartment, where it is bound by the sulfide-binding hemoglobins present in the blood and transported to the bacterial symbionts. H₂S movement into these worms, however, appears to be severely limited.

Both freshly collected and experimental worms had higher internal ΣH₂S levels as the surrounding water pH increased. This suggests that external HS⁻ is the most important factor affecting internal ΣH₂S values because HS⁻ is more abundant than H₂S at higher pH. In addition, at higher extracellular pH in experimental worms, the internal ΣH₂S levels were also higher, suggesting that the species of sulfide being accumulated increases the internal pH. Both multiple regressions indicate that internal ΣH₂S levels and extracellular pH are influenced more by external [HS⁻] than by external [H₂S]. This suggests that external HS⁻ levels play a greater role in the uptake of sulfide in these animals than do H₂S levels.

Our experiments involving depressed extracellular pH also support the contention that HS⁻ is the primary species of sulfide moving into the worms. We observed a different pH-dependence for sulfide acquisition from that expected for H₂S diffusion and measured for CO₂ diffusion (Goffredi et al. 1997). As the extracellular pH in these animals was depressed (by exposure to hypoxic conditions or N-ethylmaleimide), no decrease in internal sulfide levels was observed. NEM inhibits enzymes by forming covalent bonds with sulfhydryl groups (SH⁻), causing deleterious conformational changes in these enzymes (Stone et al. 1984; Lin and Randall, 1993). Thus, it is possible for NEM to react with a variety of enzymes and proteins possessing reactive sulfhydryl groups, including R. pachyptila hemoglobins, which have been shown to contain free cysteine residues (Zal et al. 1997). Experiments were conducted to determine whether NEM adversely affects sulfide binding by R. pachyptila hemoglobin (Zal et al. 1997). In summary, when R. pachyptila hemoglobin was pre-treated with NEM, prior to any exposure to sulfide, a 30 % decrease in sulfide binding resulted. However, if the hemoglobin was first exposed to sulfide, as in the case of our experiments, there was no effect of NEM on sulfide binding (Zal et al. 1997). The worms were given NEM and sulfide simultaneously; therefore, they were not pretreated with the inhibitor as sulfide would be expected to move into the worms faster than NEM. In addition, we feel that because there was no decrease in internal sulfide levels, adverse effects on other proteins did not create artifacts. Our results also show that there is discrimination against H₂S movement into the extracellular fluids of these animals. In general, organisms are believed to be unable to block the diffusion of H₂S across membranes while still retaining permeability to other gases, such as CO₂ and O₂, both of which diffuse into R. pachyptila (Somero et al. 1989; Bagarinao, 1992; Völkel, 1995; Goffredi et al. 1997). For example, it has been shown that the shrimp Crangon crangon, which inhabits shallow sandy areas, is only permeable to H₂S and that there is no uptake of HS⁻ (Vismann, 1996). Researchers have also shown that sulfide penetration into the alga Valonia macrophysa increased as external pH decreased, indicating that for this alga H₂S is the more permeable of the two sulfide species (Jacques, 1936). Although Urechis caupo, the fat innkeeper worm, shows a higher H₂S permeability, HS⁻ permeability across the body wall has been demonstrated to be 37 % of the H₂S permeability (Julian and Arp, 1992). In contrast, it appears that H₂S movement into R. pachyptila is much lower than expected, limited by some currently unknown mechanism.

The specific mechanism for HS⁻ uptake is also unknown at this time; however, we propose that HS⁻ enters via facilitated diffusion due to its charge and because of the strong correlation between internal ΣH₂S and external [HS⁻]. Although [HS⁻] is relatively low in the vent environment (50 μmol l⁻¹ at pH 6.0 and 300 μmol l⁻¹ ΣH₂S), a gradient for HS⁻ movement into the worms is created and maintained by the high concentrations of sulfide-binding hemoglobins present in the body fluids (Arp and Childress, 1983; Arp et al. 1987; Fisher et al. 1988). Sulfide binding by the hemoglobins has been shown to be maximal at pH 7.5, which suggests that the actual species of sulfide bound by the hemoglobins is HS⁻ (Childress et al. 1984).

For two reasons, we believe that HS⁻ uptake, as the mechanism for sulfide acquisition in R. pachyptila, acts as a protection against sulfide poisoning. The first is the fact that perhaps the two species of sulfide ([S²⁻] is negligible, with a pK of 12–13) are not equally toxic to the animal (Bagarinao and Vetter, 1990). It has been suggested that H₂S is more toxic than HS⁻, and that H₂S is actually the species of sulfide that binds to the cytochrome c oxidase complex (Smith et al. 1977; Powell and Somero, 1986; Bagarinao and Vetter, 1990; Oeschger and Vismann, 1994). Specifically, Powell and Somero (1986) have shown that cytochrome c oxidase activity in R. pachyptila plume tissue, in the presence of sulfide, decreases markedly with decreasing extracellular pH (from 7.0 to 6.0), suggesting that H₂S is the more inhibitory form of sulfide. In a similar experiment, it was shown that there was no HS⁻ inhibition of mitochondrial respiration of the killifish Fundulus parvipinnis and, again, that H₂S was the toxic form of sulfide (Bagarinao and Vetter, 1990). If this were the case for R. pachyptila, it would be advantageous for these worms to exclude H₂S while importing HS⁻.

The second possibility is that if inorganic carbon and sulfide were acquired via the same mechanism, i.e. diffusion of the undissociated form (CO₂ and H₂S), R. pachyptila could not control sulfide uptake independently. Thus, the second way in which HS⁻ uptake could protect R. pachyptila from sulfide poisoning would be to decouple sulfide acquisition from inorganic carbon acquisition. Although R. pachyptila must maintain an alkaline extracellular pH in order to concentrate inorganic carbon internally, if sulfide were accumulated in the same way, via H₂S diffusion, free sulfide, like ΣCO₂, would reach very high concentrations in the blood. This unlimited accumulation of sulfide internally could potentially poison the worm and its symbionts. However, with the proposed mechanism, sulfide uptake is expected to be largely limited by the binding capacity of the hemoglobins.

In contrast, the vesicomyid clam Calyptogena elongata, which also contains chemoautotrophic symbionts that it must supply with sulfide, does not maintain its extracellular pH constant in the face of changing external and internal ΣH²S levels (Childress et al. 1993a). Specifically, when exposed to increasing amounts of internal ΣH₂S, the extracellular pH of C. elongata decreases (Childress et al. 1993a). C. elongata can potentially accumulate sulfide via H₂S diffusion because increasing levels of internal ΣH₂S cause the extracellular pH to decrease, which dissipates the sulfide gradient into the animal, acting as a self-limiting sulfide acquisition mechanism for the clam.

R. pachyptila, however, does not rely on the same self-limiting mechanism and has apparently evolved an alternative mode of sulfide acquisition, mediated transport of HS−, as well as reduced permeability to, or some discrimination against, H₂S, apparently as a protection against sulfide poisoning. In this way, R. pachyptila is able to control sulfide movement, while keeping the extracellular pH stable and alkaline. Restricting the internal ΣH₂S level to that which can be bound by the hemoglobins ensures that, even at high external ΣH₂S levels, internal sulfide levels in R. pachyptila are not toxic to either partner but are still sufficient for the symbionts. This mechanism appears to be a further specialization of R. pachyptila for successfully supporting autotrophic endosymbionts and thriving in such a hostile environment.

Funding for this project was provided by NSF grants OCE-9301374 (J.J.C.) and OCE-9632861 (J.J.C.) and by Ifremer URM 7 (F.H.L.). The authors thank the captains and crew of the R.V. New Horizon, R.V. Atlantis II, R.V. Nadir, R.V. Wecoma, D.S.R.V. Alvin and D.S.R.V. Nautile. We also thank

R. Lee, P. Girguis, S. Powell, J. Freytag and J. Smith for technical support at sea. Finally, we would like to thank L. Mullineaux and F. Gaill, chief scientists on the HERO 94 and HOT 96 expeditions, respectively.