Sulphaemoglobin Formation in Fish: A Comparison Between the Haemoglobin Of the Sulphide-Sensitive Rainbow Trout (Oncorhynchus Mykiss) and of the Sulphide-Tolerant Common Carp (Cyprinus Carpio)

Toxic sulphide can accumulate in aquatic environments when organic material is decomposed under oxygen-limiting conditions (Nedwell, 1982). (The term sulphide is intended to include undissociated H₂S, the bisulphide anion HS⁻ and the sulphide anion S²⁻.) In marine sediments, sulphide production is enhanced by bacterial sulphate reduction (Jorgensen and Fenchel, 1974; Nedwell, 1982). Species inhabiting intertidal flats or salt marshes are, therefore, frequently exposed to considerable concentrations of sulphide (Bagarinao, 1992; Grieshaber and Völkel, 1998). When the sulphide concentration in the sediment is high, sulphide can reach significant concentrations in the water column and may affect, for example, fishes living in shallow waters along the coast (Bagarinao, 1992). Tidal marsh inhabitants such as the California killifish (Fundulus parvipinnis) and the long-jawed mudsucker (Gillichthys mirabilis) exhibit very high sulphide tolerances, with 96 h LC₅₀ values of 700 and 525 μmol l⁻¹, respectively. In contrast, active swimmers and open coast inhabitants such as the speckled sanddab (Citharichthys stigmaeus) show little sulphide tolerance (Bagarinao and Vetter, 1989). Sulphide concentrations are generally lower in freshwater systems than in marine habitats. Nevertheless, sulphide can accumulate in the hypolimnion of eutrophic lakes and in fish farms (Reynolds and Haines, 1980; Bark and Goodfellow, 1985). Many freshwater fishes, including the rainbow trout (Oncorhynchus mykiss), are sensitive to sulphide, showing 96 h LC₅₀ values in the low micromolar range (Smith et al., 1976). Some species, however, exhibit strikingly high sulphide tolerances. The common carp (Cyprinus carpio), for example, survives in highly polluted river water and is able to withstand sulphide concentrations of approximately 300 μmol l⁻¹ for 1 month (Kumar and Mukherjee, 1988).

Membranes are in general highly permeable to sulphide (Beermann, 1924) and it is, therefore, very likely that sulphide quickly penetrates the gill epithelium of fish. Sulphide entering the blood may interact with haemoglobin (Hb) by forming sulphaemoglobin (SHb), a green Hb derivative with covalently bound sulphur at the pyrrole of the porphyrin ring (Morell et al., 1967; National Research Council, 1979). Ferrous (FeII) SHb binds O₂ with a much lower affinity than Hb: the P₅₀ of SHb is approximately two orders of magnitudes higher than the P₅₀ of unaltered Hb (Carrico et al., 1978b). Therefore, SHb is practically non-functional in O₂ transport. Since SHb is not reconverted to Hb within a reasonable time frame under physiological conditions (Johnson, 1970; Carrico et al., 1978a; Bagarinao and Vetter, 1992), pronounced SHb formation may seriously impair O₂ provision in fish.

On the formation of SHb, the optical spectrum of Hb changes drastically, showing a new absorption maximum at approximately 617–622 nm (Drabkin and Austin, 1935a; Nichol and Morell, 1968; Dijkhuizen et al., 1977). The quantitative determination of SHb, however, is difficult, because until now it has not been possible to produce pure SHb (Siggaard-Andersen et al., 1972; Dijkhuizen et al., 1977; Carrico et al., 1978a). Spectra of pure SHb can, therefore, be obtained only by calculation. This has been done, for example, by quantitative determination of the SHb content of a SHb/HbO₂ mixture by electron paramagnetic resonance spectra (Carrico et al., 1978a) or by determination of sulphide-induced loss of O₂-binding capacity (Dijkhuizen et al., 1977). In many studies, the concentration of SHb in blood was determined using the extinction coefficient at its absorption maximum (617–622 nm; Zwart et al., 1981; Brittain et al., 1982; Park et al., 1986). In other cases, the absorbance ratio A_620nm/A₅₈₀ nm (or at other very close wavelengths) was used to determine the purity of an SHb mixture (Nichol et al., 1968; Johnson, 1970; Brittain, 1981; Bagarinao and Vetter, 1992). Both methods, however, can be problematic since the absorbances at 620 and 580 nm change not only during the formation of SHb but also during deoxygenation or the formation of methaemoglobin (see, for example, Benesch et al., 1973; Salvati and Tentori, 1981).

The question of whether SHb formation may contribute to sulphide poisoning has long been a matter of debate (Haggard, 1925; National Research Council, 1979; Curry and Gerkin, 1987). SHb formation in fish blood was investigated by Bagarinao and Vetter (1992). They found only small amounts of SHb in California killifish (F. parvipinnis) and mudsuckers (G. mirabilis) killed by high concentrations of sulphide (2–8 mmol l⁻¹). In vitro, SHb was formed in the haemolysate of both species only at sulphide concentrations of 1 mmol l⁻¹ or higher. The authors concluded that SHb formation is not involved in acute sulphide poisoning of F. parvipinnis and G. mirabilis (Bagarinao and Vetter, 1992). It is now generally believed that the formation of SHb is of minor importance when an organism is exposed to a lethal dose of sulphide since the acute toxicity of sulphide is mainly due to its inhibition of cytochrome c oxidase (Nicholls, 1975; National Research Council, 1979; Smith and Gosselin, 1979). It is still not known, however, whether SHb contributes to sulphide toxicity during chronic exposure to sublethal levels of sulphide. Until now, SHb formation in fish has been studied only in sulphide-tolerant species. The scope of the present paper is (i) to establish a reliable method for the quantitative determination of SHb in fish blood, (ii) to investigate SHb formation in a sulphide-sensitive fish and to compare it with that in a sulphide-tolerant fish species, (iii) to study factors that may favour SHb formation during chronic exposure to sulphide and (iv) to examine SHb formation within red blood cells.

Rainbow trout Oncorhynchus mykiss (Walbaum) (200–500 g) and common carp Cyprinus carpio L. (550–750 g) were obtained from commercial sources. Fish were maintained in large aquaria with aerated recirculating Berlin tapwater for at least 1 week at 17–21 °C in summer and 12–16 °C in winter, and with a nitrite concentration of no more than 0.05 mg l⁻¹. Pig blood was obtained from a local abattoir.

Chemicals were purchased from Merck (Darmstadt, Germany) or Sigma (Deisenhofen, Germany). Trout saline consisted of (in mmol l⁻¹) NaCl (145), NaHCO₃ (5), KCl (4), MgCl₂ (1.2), ^D-glucose (5) and CaCl₂ (1.3) (after Nikinmaa et al., 1987). In Hepes-buffered trout saline (pH 7.9 at 20 °C), the concentration of NaCl was changed to 80 mmol l⁻¹, and 40 mmol l⁻¹ Hepes was added. Carp saline contained (in mmol l⁻¹) NaCl (128), NaHCO₃ (5), KCl (5), MgCl₂ (1.5), D-glucose (5) and CaCl₂ (1.5) (modified from Salama and Nikinmaa, 1988). In Hepes-buffered (40 mmol l⁻¹) carp saline, the NaCl concentration was 93 mmol l⁻¹, with concentrations of all other components remaining unchanged (pH 7.9, 20 °C). Sulphide stock solution (200 mmol l⁻¹) was made from washed crystals of Na₂S.9H₂O (Merck) diluted in O₂-free distilled water. Modified Drabkin’s reagent consisted of NaHCO₃ (11.9 mmol l⁻¹), K₃[Fe(CN)₆] (0.61 mmol l⁻¹) and KCN (0.77 mmol l⁻¹) (Tentori and Salvati, 1981).

Sulphide concentrations were determined by high-performance liquid chromatography (HPLC) after derivatization with monobromobimane (Fahey et al., 1981; Newton et al. 1981; Vetter et al., 1989) as described by Völkel and Grieshaber (1992). A Hewlett Packard HPLC series 1050 (Hewlett Packard, Waldbronn, Germany) combined with an HP LiChrospher 60 RP select B (5 μm) 125-4 column was used. Thiols were detected using an HP 1046A fluorescence detector (excitation wavelength 380 nm, emission wavelength 480 nm).

Fish were killed by a sharp blow to the head, and blood was obtained by caudal venipuncture using heparinized hypodermic syringes. Red blood cells were washed three times in 3–5 volumes of ice-cold trout or carp saline, respectively, each time removing the buffy coat. The resulting red cell suspensions were oxygenated and incubated overnight without mixing at 6 °C. The red blood cells were washed again and suspended to their original haematocrit (Hct 20–30 %; Compur M 1100, München, Germany) and were then haemolysed by freezing at −70 °C. Prior to the experiments, the suspensions were thawed and centrifuged for 5 min at 5 °C (20 200 g, Eppendorf centrifuge 5417R, Eppendorf, Hamburg, Germany) to pellet the cell debris. The haem content of the clear haemolysates was determined.

In the context of this paper, all Hb concentrations are expressed as mmol haem l⁻¹. Total Hb concentration ([Hb]_tot) was measured according to a modified method described by Tentori and Salvati (1981). Drabkin’s reagent (1 ml) was mixed with 0.5 ml of saponin (0.05 %) and 20 μl of haemolysate. A millimolar extinction coefficient for methaemoglobincyanide at 540 nm of 11 l cm⁻¹ mmol⁻¹ was used (Tentori and Salvati, 1981). In solutions containing SHb, the Fe content of the sample was measured using a Perkin-Elmer 2380 atomic absorption spectrophotometer (Perkin-Elmer, Überlingen, Germany) with 1 mol of Fe taken to be equivalent to 1 mol of Hb monomer (van Kampen and Zijlstra, 1965). The accuracy of this method was checked using fresh SHb-free trout haemolysate giving a millimolar extinction coefficient for methaemoglobincyanide at 540 nm of 11.03±0.41 l cm⁻¹ mmol⁻¹ (mean ± S.D., N=3).

Spectral scans were made from 480 to 700 nm in 1 nm steps using a temperature-controlled Beckman DU70 spectrophotometer (Beckman Instruments, Fullerton, CA, USA) together with software designed in our department using the TestPoint program (Keithley Instruments, Germering, Germany). All determinations were performed in a half micro-cuvette of 1 cm path length.

To validate the spectral deconvolution technique described above, blood from a pig was haemolyzed, and the Hb content was determined as described above. The haemolysate was diluted to a [Hb]_tot of approximately 60 μmol l⁻¹ with 50 mmol l⁻¹ citrate and 100 mmol l⁻¹ KCl (pH 5.5). Samples of K₃[Fe(CN)₆] (0.6–60 μmol l⁻¹ final concentration) were added to obtain equimolar concentrations of methaemoglobin. The concentration of the K₃[Fe(CN)₆] stock solution was checked by measuring its absorbance at 420 nm with water as a blank. An extinction coefficient of 1.01 l mmol⁻¹ cm⁻¹ for K₃[Fe(CN)₆] at 420 nm was used (van Kampen and Zijlstra, 1965). After a 30 min incubation at 25 °C, the spectra of the haemolysate samples were measured and analyzed. The change in [Hb⁺] (sum of acid and alkaline Hb⁺) obtained via spectral deconvolution, y (μmol l⁻¹), was plotted against the theoretical increase in [Hb+] calculated from the concentration of added K₃[Fe(CN)₆], x (μmol l⁻¹). The linear regression fit was y=1.007x−0.17 (r=0.9997, P<0.001). The inaccuracy of the spectral deconvolution technique was below 1 μmol l⁻¹. The lowest [Hb⁺] that could be determined with this accuracy was 1.2 μmol l⁻¹. The sum of all haemogobin derivatives found in these samples (HbO₂, DeoxyHb and Hb⁺) was 96.9±0.8 % (mean ± S.D., N=8) of [Hb]_tot as assessed using Drabkin’s reagent.

To test the variability of the analyses, eight subsamples of a haemolysate containing all five Hb components (83 % HbO₂, 1 % DeoxyHb, 2 % Hb⁺_acid, 3 % Hb⁺_alkaline and 11 % SHb) were diluted to 70 μmol l⁻¹ Hb, and the concentration of each compound was determined using spectral deconvolution. The concentration of each compound deviated by less than 1 μmol l⁻¹ between the samples. The imprecision, therefore, was less than 2 % of the mean [Hb]_tot used in this study.

To calculate the concentrations of each Hb component according to equation 2, it was necessary to determine the millimolar extinction coefficient of the respective component at each wavelength between 480 and 700 nm in 1 nm steps. For the preparation of pure Hb derivatives, haemolysate was obtained as described above. [Hb]_tot was measured using Drabkin’s reagent. For the preparation of pure HbO₂ and DeoxyHb, the haemolysate was diluted with air-saturated KCl/Hepes buffer (100 mmol l⁻¹/50 mmol l⁻¹) at pH 8.0 to avoid deoxygenation by the Root effect (see Pelster and Weber, 1991). Haemolysate (1 ml) was placed into a half micro-cuvette and, after 3 min of temperature equilibration (5, 10, 15, 20 or 25 °C), the pure HbO₂ spectrum was measured. A few milligrams of Na₂S₂O₄ was added and, after 1–3 min, a scan of the pure DeoxyHb was recorded. For the preparation of pure Hb⁺, the haemolysate was diluted in KCl/Hepes buffer (100 mmol l⁻¹/50 mmol l⁻¹) at pH 6.5, and K₃[Fe(CN)₆] was added in a molar ratio of 8 mol mol⁻¹ Hb monomer (Austin and Drabkin, 1935). After 30 min, the reaction was completed and the Hb⁺ stock solution was diluted 10-fold in KCl/Hepes buffer at pH 6.5 and 9. The pH in the haemolysate was 6.22±0.01 and 8.65±0.04, respectively (15 °C, means ± S.D., N=5). The spectra of the acid and alkaline Hb⁺ were recorded. The millimolar extinction coefficients of HbO₂, DeoxyHb, Hb⁺_acid and Hb⁺_alkaline were obtained by dividing the absorbance at each wavelength by the millimolar Hb concentration.

To determine the extinction coefficients of SHb, a modified method after Siggaard-Andersen et al. (1972) was used. As mentioned above, SHb cannot be obtained in pure form. The extinction coefficients, therefore, have to be calculated from the spectra of Hb solutions containing a large fraction of SHb. For this purpose, intact red blood cells were used instead of haemolysate to avoid denaturation of Hb and the accumulation of methaemoglobin (see below). Blood from trout was washed and incubated overnight as described above. The red blood cell solution was adjusted to a high haematocrit (30–40 %) using Hepes-buffered trout saline, and the cells were oxygenated in a glass tonometer (Eschweiler, Kiel, Germany) for 20 min with humidified air at 15 °C. The red cells were then placed into an air-tight flask, and neutralized sulphide stock solution was added to give a final concentration of 50 mmol l⁻¹. The cells were incubated for 1 h at 15 °C. The cell suspension was then centrifuged (2 min, 110 g, 15 °C), and the supernatant was exchanged with fresh Hepes-buffered saline to remove excess sulphide. After 30 min of reoxygenation, the cells were again incubated with sulphide. The whole procedure was repeated three times. After the last incubation, the cells were washed in Hepes-buffered saline, reoxygenated and then haemolysed as described above. [Hb]_tot was determined, and the spectrum was measured after dilution in KCl/Hepes buffer (pH 7.5) as described above. A few milligrams of Na₂S₂O₄ was then added and, after 1–3 min, the spectrum of the deoxygenated sample was measured. The fraction of SHb in the Hb solution and the extinction coefficients of SHb between 480 and 700 nm were calculated after Siggaard-Andersen et al. (1972) using only solutions with an SHb content of 20 % or more.

The extinction coefficients of HbO₂, DeoxyHb, Hb⁺_acid, Hb⁺_alkaline and SHb from trout at 15 °C are shown in Fig. 2. The millimolar extinction coefficients at the respective wavelength maxima at 15 °C of trout and carp Hb are given in Table 1. The wavelength maxima and extinction coefficients at 5–25 °C did not differ significantly from the values at 15 °C (data not shown).

Sulphaemoglobin formation was measured in the haemolysate of trout and carp in five series of experiments. The haemolysate was diluted with KCl/Hepes (100 mmol l⁻¹/50 mmol l⁻¹, pH adjusted at room temperature, 22 °C; see experimental conditions for measured pH in the haemolysate). [Hb]_tot was 51.8±6.4 μmol l⁻¹ (trout) and 51.3±10.9 μmol l⁻¹ (carp) (means ± S.D., N=22), apart from series 5 (see below). Haemolysate (0.995 ml) was placed into a cuvette and, after 3 min of temperature equilibration, a control scan was recorded. KCl/Hepes buffer with the respective pH and temperature was used as a blank. Sulphide solution (5 μl) (the concentration of which depended on the desired initial sulphide concentration) was added. In the following, sulphide concentrations are given as initial concentrations.

Series 1: dependence on time

Haemolysate was diluted in KCl/Hepes buffer at pH 7.5. The haemolysate was equilibrated at 15 °C, and a control spectrum was measured. Sulphide (1 mmol l⁻¹) was added, and spectra were recorded at 2.5 min intervals (0–20 min) followed by 5 min intervals (20–60 min) (N=6 at time points 0, 5, 15, 30 and 60 min, N=3 at all other time points).

Series 2: dependence on sulphide concentration

Spectra were recorded before (control) and 15 min after the addition of 0.1 mmol l⁻¹ sulphide. The same procedure was followed with sulphide concentrations of 0.25, 0.5, 0.75 and 1 mmol l⁻¹ (N=6, pH 7.5, 15 °C).

Series 3: dependence on temperature

Haemolysate was equilibrated at 5 °C, and spectra were recorded before (control) and 15 min after addition of 1 mmol l⁻¹ sulphide. The experiment was repeated at 10, 15, 20 and 25 °C (N=5–6, pH 7.5).

Series 4: dependence on pH

Haemolysate was diluted in KCl/Hepes buffer at pH 6.5. Spectra were recorded before (control) and 15 min after addition of 1 mmol l⁻¹ sulphide. The experiment was repeated with haemolysate diluted in KCl/Hepes buffer at pH values of 7.0, 7.5, 8.0 and 8.5 (N=6, 15 °C).

Series 5: dependence on haemoglobin concentration

Haemolysate was diluted with KCl/Hepes buffer, yielding a final [Hb]_tot of between 15 and 180 μmol l⁻¹ (trout) and 20 and 100 μmol l⁻¹ (carp). Spectral scans were recorded before (control) and 15 min after addition of 1 mmol l⁻¹ sulphide. SHb formation was measured at various values of [Hb]_tot in haemolysate from 6–9 different animals (pH 7.5, 15 °C).

The pH of the haemolysate was checked before and after the addition of sulphide (BMS 3 MK 2, Radiometer, Copenhagen, Denmark, thermostatted at the respective temperature and calibrated using precision buffers). At 15 °C, the pH values of control haemolysates (without sulphide) diluted in KCl/Hepes buffer (pH 7.5) were 7.482±0.003 (trout) and 7.468±0.003 (carp) (means ± S.D., N=4). The pH values in the haemolysates after the addition of 1 mmol l⁻¹ sulphide were 7.533±0.004 (trout) and 7.529±0.009 (carp) (N=5). For pH values of haemolysate from series 4, see Fig. 4B. Sulphide concentrations after the addition of 1 mmol l⁻¹ sulphide were followed at 15 °C and at a [Hb]_tot of 61.0±9.0 μmol l⁻¹. Sulphide concentrations in trout haemolysates decreased from 781±20 μmol l⁻¹ at 1 min after sulphide addition to 634±50 μmol l⁻¹ after 15 min and 316±138 μmol l⁻¹ after 60 min of incubation (means ± S.D., N=3). Similar concentrations were measured in carp haemolysates (data not shown). Under the same conditions, sulphide concentrations in controls (buffer alone) dropped to 719±171 μmol l⁻¹ after 60 min.

Blood from trout and carp was washed and incubated overnight as described above. The red blood cell suspension was adjusted to a haematocrit of approximately 30 % using Hepes-buffered saline. The cells were oxygenated in a tonometer as described above and were then incubated in an air-tight flask either without (controls) or with 2.5 mmol l⁻¹ sulphide for 60 min at 15 °C. Sulphide concentrations were determined 5 min and 60 min after the addition of sulphide, and pH was measured every 15 min. At the end of the incubation, excess sulphide was removed, as described above, and the cells were oxygenated for at least 30 min. Samples were then taken for the determination of [Hb]_tot, sulphide and SHb content. The oxygen content of the red blood cell suspensions was measured using the method of Tucker (1967) as modified by Bridges (1983).

In these experiments, sulphide concentrations within the red blood cell suspensions dropped from 1.01±0.13 mmol l⁻¹ after 5 min to 0.24±0.12 after 60 min in trout and from

1.26±0.09 mmol l⁻¹ after 5 min to 0.30±0.05 mmol l⁻¹ in carp (means ± S.D., N=4). Sulphide concentrations were below 10 μmol l⁻¹ after reoxygenation. The pH values in the red blood cell suspensions were 8.137±0.066 in trout and 8.238±0.037 in carp (means ± S.D., N=5–7).

Unless otherwise stated, values are given as means ± standard deviation (mean ± S.D.) of the results from N experiments with blood from different animals. SHb fractions are expressed as mol SHb mol⁻¹ Hb_tot. Differences between means were evaluated using a statistical software package (SigmaStat, Jandel Scientific, San Rafael, CA, USA). For a pairwise comparison between values for carp and trout, the Mann–Whitney rank sum test was used at the P<0.01 level. Differences between the treatments within one species were evaluated using a one-way analysis of variance (ANOVA) on ranks using Dunnett’s method at the P<0.05 level.

On the addition of sulphide, the absorption spectra of trout and carp Hb changed drastically. Fig. 3A shows the absorption spectra of trout Hb before and after the addition of 1 mmol l⁻¹ sulphide. Within 5 min of incubation, a shoulder at 618 nm appeared and the typical absorption maxima of HbO₂ decreased. After 15 min, a distinct absorption peak had developed at 618 nm, and this increased further over the next 45 min. At the same time, the HbO₂ peaks at 575 and 540 nm decreased and had almost disappeared after 60 min. The spectral changes were converted into the fraction of total Hb present as sulphaemoglobin (Fig. 3B). SHb formation showed a sigmoidal time course in the haemolysate of both animals, levelling off at an SHb fraction of approximately 0.1 mol SHb mol⁻¹ Hb_tot in carp and at approximately 0.3 mol SHb mol⁻¹ Hb_tot in trout. The first derivative of the sigmoidal curve was used to determine the time when the rate of SHb formation was maximal (11 min in trout and 8 min in carp haemolysate). The rates of SHb formation at these times were 0.007 mol SHb mol⁻¹ Hb min⁻¹ in trout haemolysate and 0.004 mol SHb mol⁻¹ Hb min⁻¹ in carp haemolysate.

In the haemolysate of both animals, the SHb fraction increased with increasing sulphide concentration (Fig. 4A). In trout Hb, the SHb fraction after 15 min of sulphide incubation increased almost linearly from approximately zero (0.1 mmol l⁻¹) to 0.13±0.02 mol SHb mol⁻¹ Hb_tot at 1 mmol l⁻¹ sulphide (N=6). A linear, although much lower, increase in SHb fraction was observed in carp haemolysate between 0.25 and 1 mmol l⁻¹ sulphide (from zero to 0.06± 0.01 mol SHb mol⁻¹ Hb_tot (N=6).

In addition, SHb formation was markedly influenced by pH (Fig. 4B). In trout haemolysate, the SHb fraction after 15 min of sulphide incubation (1 mmol l⁻¹ sulphide) was only approximately zero at pH 8.7 (N=6). A decrease in pH to 8.1 caused a slight but significant increase in SHb concentration. A further decrease in pH, however, caused a marked increase in the size of the SHb fraction to 0.22±0.02 mol SHb mol⁻¹ Hb_tot at pH 7.1 (N=6). Lowering the pH to 6.8 had no further effect on the SHb fraction in trout haemolysate. In carp haemolysate, the SHb fraction increased almost linearly from approximately zero at pH 7.9 to 0.15±0.01 mol SHb mol⁻¹ Hb_tot at pH 6.8 (N=6).

Temperature had a profound effect on SHb formation in trout haemolysate (Fig. 4C). The SHb fraction after 15 min (1 mmol l⁻¹ sulphide) was only 0.03±0.02 mol SHb mol⁻¹ Hb_tot at 5 °C, but increased significantly to 0.28± 0.02 mol SHb mol⁻¹ Hb_tot at 25 °C (N=5). The temperature effect was more pronounced at lower temperatures, with a Q₁₀ value of 3.8±1.1 between 5 and 15 °C and a Q₁₀ of 2.2±0.3 between 15 and 25 °C (N=5). In carp haemolysate, the SHb fraction after 15 min was 0.03±0.01 mol SHb mol⁻¹ Hb_tot at 5 °C and 0.06±0.01 mol SHb mol⁻¹ Hb_tot at 25 °C (N=6). Q₁₀ values remained between 1.0 and 1.4 between 5 and 25 °C.

In both carp and trout haemolysate, SHb concentration was linearly related to Hb concentration (Fig. 5). From the slope of the linear regression fit, it can be estimated that, in trout haemolysate, 0.15 mol SHb mol⁻¹ Hb was produced during 15 min of sulphide incubation (1 mmol l⁻¹ sulphide), whereas under the same conditions only 0.06 mol SHb mol⁻¹ Hb was produced in carp haemolysate. These values are in good agreement with the rate of SHb formation calculated from Fig. 3B.

To measure SHb formation and its influence on Hb oxygen-saturation within intact red blood cells of trout and carp, red blood cell suspensions were incubated with 2.5 mmol l⁻¹ sulphide. After 60 min of incubation, the fraction of SHb within trout red blood cells amounted to 0.07± 0.02 mol SHb mol⁻¹ Hb_tot (N=5). Under the same conditions, the Hb oxygen-saturation (oxygen bound per Hb monomer) was significantly reduced by 22 % from 0.9±0.1 in controls to 0.7±0.1 after sulphide incubation (Fig. 6). In carp red blood cells, sulphide incubation induced the formation of only 0.01±0.01 mol SHb mol⁻¹ Hb (N=7) and Hb oxygen-saturation was not significantly affected (0.9±0.1 in controls and in sulphide incubations).

In the present study, the quantitative determination of SHb in mixtures of Hb derivatives was performed using the spectral deconvolution technique. A prerequisite for using this method is the availability of extinction coefficients for the components involved over the relevant spectral range. The spectra of HbO₂ and DeoxyHb from trout and carp exhibited similar absorption maxima and millimolar extinction coefficients to those measured for mammalian Hb (see, for example, van Kampen and Zijlstra, 1965; Salvati and Tentori, 1981; Zwart et al., 1981). It appeared to be very important, however, to measure the spectra of Hb⁺_acid and Hb⁺_alkaline under the same conditions as in the experimental assays because the spectrum of methaemoglobin depends on factors such as pH, ionic strength and temperature (Austin and Drabkin, 1935; Benesch et al., 1973). In contrast to the other Hb derivatives mentioned above, a laborious method is necessary to obtain the spectrum of pure SHb. The first step was the production of Hb solutions with high fractions of SHb. The highest achievable fraction of SHb was approximately 35 % in trout red blood cells. Carp erythrocytes yielded only minimal amounts of SHb and could, therefore, not be used for the production of SHb.

During the repeated sulphide incubations and oxygenation periods, small amounts of methaemoglobin may be formed which may then react with sulphide to form ferric (FeIII) sulphaemoglobin. The methaemoglobin content and the concentration of ferric sulphaemoglobin in our experiments, however, seemed to be negligible because the final spectrum did not change upon the addition of cyanide (Evelyn and Tait Malloy, 1938; Robin and Harley, 1964). This may be because methaemoglobin was produced in only small amounts. Siggaard-Andersen et al. (1972) state that methaemoglobin formation is minimal when the molar ratio of sulphide added to the concentration of Hb is approximately 2:1. In our experiments, the nominal molar ratio within the red blood cells was approximately 3:1, but the actual sulphide concentration may have been somewhat lower as a result of diffusive losses. Another possibility is that any methaemoglobin formed was quickly reduced by the intact erythrocytes (Freeman et al., 1983). For a discussion of other sources of error, see Siggaard-Andersen et al. (1972).

The computed spectrum of trout SHb obtained in the present study (Fig. 2) is very similar to mammalian SHb spectra (e.g. Nichol and Morell, 1968; Carrico et al., 1978a), although the absorption between 590 and 500 nm is somewhat higher in trout SHb. The millimolar extinction coefficient at the absorption maximum (618 nm) of 21.1±0.3 l mmol⁻¹ cm⁻¹ is close to the values obtained in several other studies on mammals (between 20.8 and 21.6 l mmol⁻¹ cm⁻¹; Nichol and Morell, 1968; Dijkhuizen et al., 1977; Carrico et al., 1978a; Salvati and Tentori, 1981). The use of trout SHb extinction coefficients for the analysis of carp Hb spectra may be problematic. The absorption maximum of carp SHb, however, is very close to the respective maximum of trout SHb (Table 1). This, together with the fact that the trout extinction coefficient of SHb at 618–620 nm is very similar to that of other species, indicates that the trout SHb spectrum can be applied to the spectral analysis of carp Hb without causing a major error.

Another source of error during spectral analysis of Hb mixtures is the possible denaturation of Hb, which can occur at sulphide concentrations in the millimolar range and during long-term incubations (Bagarinao and Vetter, 1992). Denaturation of Hb causes the formation of haemi- and haemochromes. The spectral changes due to the denaturation of Hb are difficult to assess. Drabkin and Austin (1935b) give various spectra of haemochromes under different experimental conditions. Both haemi- and haemochromes appear to exhibit absorption spectra with two peaks at approximately 535 and 565 nm, and total absorption increases between 700 and 480 nm (Drabkin and Austin, 1935b; Di Iorio, 1981). In addition to the denaturation of Hb, any Hb⁺ present in the samples may react with sulphide, forming ferric SHb (Brittain et al., 1982). However, [Hb⁺] was negligible in most of the samples, and the spectra did not change on the addition of cyanide, indicating that [SHb⁺] was also minimal. In some samples, Hb⁺ was present, but the concentration never exceeded 5 % of [Hb]_tot.

The presence of absorbing components other than the Hb derivatives examined can lead to erroneous calculations of the respective Hb derivative concentrations. These errors, however, are easily detected during spectral analysis because, in this case, the calculated spectrum starts to deviate increasingly from the measured spectrum. Moreover, the presence of interfering components causes drastic changes in isosbestic points. In the present study, Hb denaturation was negligible in controls and during sulphide incubations up to 1 mmol l⁻¹ and up to 30 min, but it was favoured at low pH values and high temperatures. In this case, the spectrum of diluted milk, which exhibits increasing absorption between 700 and 480 nm, was used as a sixth component for the nonlinear regression fit. Using this procedure, the coefficients of variation of the variables were reduced by approximately twofold. For example, in a sample of trout Hb (1 mmol l⁻¹ sulphide, 60 min, pH 7.5, 15 °C), the coefficient of variation for [SHb] decreased from 4.0 to 2.2 %. This was not due to the inclusion of a sixth component per se because, when the mirror spectrum of diluted milk was used instead of the milk spectrum, the value increased to 5.4 %.

However, the denaturation of Hb and formation of SHb⁺ must be regarded as possible sources of error, especially under extreme assay conditions. Denaturation may also be problematic during the generation of high SHb fractions in red blood cells for the calculation of the SHb extinction coefficients. The standard deviation of the computed SHb extinction coefficient is quite small around the absorption maximum (approximately 1 %) but is higher at other wavelengths, especially in the range 520–590 nm, which may be caused by the formation of haemo-or haemichromes. To test whether these variations are a major source of error, several Hb solutions were analyzed. Calculations using the highest and lowest values of the extinction coefficients at 520–590 nm revealed variations between the calculated SHb fractions of less than 0.2 %.

Significant sulphaemoglobin formation was observed in haemolysates of the sulphide-sensitive rainbow trout and, to a lesser extent, in haemolysates of the sulphide-tolerant common carp. At 5 min after the addition of 1 mmol l⁻¹ sulphide, the SHb fraction amounted to approximately 2 % in haemolysate of both animals, and this increased to approximately 30 and 10 % in trout and carp haemolysate, respectively, after 60 min of incubation (Fig. 3B).

The sigmoidal time course of SHb formation in the haemolysate of both animals may indicate that an autocatalytic process is involved. SHb formation in fish Hb has been studied so far only in the sulphide-tolerant species Fundulus parvipinnis and Gillichthys mirabilis (Bagarinao and Vetter, 1992). SHb formation was assessed by the absorbance ratio A₆₁₈/A₅₇₆. In the haemolysate of F. parvipinnis, the ratio increased almost linearly from zero to approximately 0.4 after 45 min of incubation with 1 mmol l⁻¹ sulphide (Bagarinao and Vetter, 1992). To compare the data from trout and carp with the data from F. parvipinnis, the absorbance ratio was calculated in the same manner as described in the study of Bagarinao and Vetter (1992). At 45 min after the addition of 1 mmol l⁻¹ sulphide, A₆₁₈/A₅₇₆ increased to 0.24±0.02 in carp haemolysate (N=3). This value is even lower than the SHb fraction in F. parvipinnis. In contrast, A₆₁₈/A₅₇₆ in trout haemolysate was much higher (1.03±0.16, N=3). Using the absorbance ratio A₆₁₈/A₅₇₆, however, the SHb fraction may be overestimated because Hb is deoxygenated during exposure to sulphide. After 45 min of incubation at 1 mmol l⁻¹ sulphide, for example, approximately 10 % of the Hb was converted to DeoxyHb, thereby increasing the absorption at 618 nm and decreasing the absorption at 576 nm.

The fact that the rate of SHb formation in trout and carp haemolysate slowed down after 20–30 min may be due (i) to the decrease in sulphide concentration in the assay, (ii) to a reconversion of SHb to Hb and (iii) to the limited availability of Hb compounds accessible to SHb formation. Fig. 4A shows that the SHb fraction depended on the sulphide concentration. In trout and carp haemolysate, the SHb fraction increased linearly between 0.1 and 1 mmol l⁻¹ in trout and between 0.25 and 1 mmol l⁻¹ in carp. These data are consistent with the findings of Bagarinao and Vetter (1992), who demonstrated that the SHb fraction increased with increasing sulphide concentration between 1 and 5 mmol l⁻¹. Similar results were obtained for dog Hb (Michel, 1938). Decreasing sulphide concentrations in the assay could, therefore, be responsible for a decrease in the rate of SHb production. In some preliminary experiments, however, sulphide concentrations as high as 5 mmol l⁻¹ were added to the haemolysate of both species. Under these conditions, SHb fractions were 38 % in trout and 10 % in carp haemolysate after 60 min of incubation (data not shown). These values are similar to the maximal SHb fractions obtained at 1 mmol l⁻¹ sulphide. It is, therefore, unlikely that a lack of sulphide was limiting SHb formation during the saturation phase shown in Fig. 3B. It is also unlikely that reconversion could prevent further accumulation of SHb since SHb formation is practically irreversible (Johnson, 1970; Carrico et al., 1978a; Bagarinao and Vetter, 1992).

Another explanation for the saturating time course of SHb formation may be the limited availability of Hb compounds accessible to SHb formation. Teleost Hbs exhibit a marked multiplicity of molecular structure and function (for a review, see Jensen et al., 1998). Within one species, they often can be electrophoretically differentiated into cathodic components with high O₂ affinities and small Bohr effects and anodic components with lower O₂ affinities and larger sensitivites to pH (Gillen and Riggs, 1973; Weber et al., 1976a,b). Pelster and Weber (1990) found that, in trout Hb, 22 % was cathodic, 66 % was anodic and 12 % was intermediate. In contrast to the anodic component, the cathodic and intermediate components (together making up 35 % of total Hb) did not exhibit a Root effect (Pelster and Weber, 1990). In contrast, carp Hb lacks a cathodic component (Weber and DeWilde, 1976), and the functional properties of its three Hb components are similar (Gillen and Riggs, 1972; Weber and Lykkeboe, 1978). It is possible that the anodic Hb components are less sensitive to sulphide than the cathodic or intermediate components. In that case, only approximately 35 % of the trout Hb and a much lower fraction of the carp Hb were accessible to SHb formation. This would be consistent with the observation that the highest obtainable fraction of SHb in trout Hb was approximately 35 % even after repeated exposures of trout erythrocytes to excess sulphide. It is not known, however, which Hb components are responsible for the formation of up to 10 % SHb in carp. Future studies will investigate whether different Hb components exhibit different sensitivities to SHb formation.

SHb formation increased markedly at pH values below 8.1 in trout haemolysate and to a lesser extent in carp haemolysate (Fig. 4B). The pH-dependence of SHb formation has been described by several authors (Michel, 1938; Johnson, 1970; Bagarinao and Vetter, 1992). Temperature also had a profound effect on SHb formation in trout haemolysate (Fig. 4C). In contrast, carp Hb was almost insensitive to temperature changes between 5 and 25 °C. The fact that low pH and high temperature accelerate the formation of SHb is striking because both factors favour the deoxygenation of fish Hb (for a review, see Jensen et al., 1998). However, SHb can only be derived from oxygenated Hb (Nichol et al., 1968). Another explanation could be that low pH values and high temperature increase the autoxidation rate of Hb (Wallace et al., 1982; Wilson and Knowles, 1987; Jensen et al., 1998). Although the mechanism of SHb formation is far from clear, it has been proposed that ferric Hb participates in the process (Nichol et al., 1968; Carrico et al., 1978a). In that case, factors that facilitate Hb oxidation may also favour SHb formation. Hb from warm-adapted fish species appears to be less sensitive to autoxidation than Hb from cold-adapted fish at temperatures between approximately 10 and 40 °C (Wilson and Knowles, 1987). The autoxidation rate of Hb from the relatively eurythermic carp is, indeed, low between 5 and 25 °C (Jensen et al., 1998). However, the question of whether Hb autoxidation favours SHb production is still a matter of debate. Oliver and Brittain (1983), for example, found that the ease of oxidation of a variety of Hbs was associated with protection against the formation of SHb.

The formation of SHb was demonstrated not only in haemolysate but also within red blood cells of rainbow trout and common carp. After 60 min of incubation with 2.5 mmol l⁻¹ sulphide, SHb amounted to 7 % of total Hb in trout red blood cells but to only 1 % in red blood cells of carp. Just as in the haemolysate, therefore, SHb formation within red blood cells was significantly lower in carp than in trout. In red blood cells of both animals, however, SHb formation was lower than in haemolysate. Although the measured sulphide concentrations were even higher in red blood cell suspensions (approximately 1 mmol l⁻¹ after 5 min of incubation as opposed to approximately 0.7 mmol l⁻¹ after 1 min of incubation in the haemolysate; see Materials and methods), the SHb content after 60 min was more than four times lower in trout and approximately 10 times lower in carp red blood cells than in the haemolysate (Fig. 3B). These differences may be because the measured sulphide concentrations in red blood cell suspensions do not necessarily correspond to intracellular concentrations. Sulphide concentrations within carp red blood cells, for example, can remain approximately five times lower than extracellular concentrations (S. Völkel, in preparation), resulting in less SHb formation.

In addition, the intracellular milieu of red blood cells is different from the haemolysate. First, Hb concentrations within the cell are much higher than those used in the experimental assays. The mean cellular Hb content of carp and trout red blood cells is approximately 17 mmol l⁻¹ (Van Raaij et al., 1996; Nielsen and Lykkeboe, 1992). The SHb fraction depended linearly on the concentration of Hb (Fig. 5). If we assume that the linear relationship can be extrapolated to high Hb concentrations, the fraction of SHb per total Hb formed within a certain time should remain constant. The higher Hb concentrations, therefore, cannot readily explain lower SHb levels within the red blood cells compared with the haemolysate.

Another factor strongly affecting SHb formation is pH (Fig. 4B). Extracellular pH values in the red blood cell suspensions during the sulphide incubations were approximately 8.1–8.2 (see Materials and methods). Under these conditions, the intracellular pH of red blood cells can be estimated to be approximately 7.8 (Tetens and Christensen, 1987; Van Raaij et al., 1996). This value is higher than the pH in the respective haemolysate experiments (approximately 7.5, see Materials and methods) and may, therefore, be partly responsible for reduced SHb formation within red blood cells. Finally, other factors such as organic phosphates or divalent cations may influence SHb formation within the red blood cells of fish.

The Hb oxygen-saturation of trout red blood cells containing approximately 7 % SHb was significantly reduced by approximately 20 % (Fig. 6). This is because that SHb is unable to bind O₂ at values around normoxia (Dijkhuizen et al., 1977; Carrico et al., 1978b). The ratio between the amount of SHb formed and the loss of oxygen saturation is, therefore, approximately 1:3. Previous studies assumed that the two values should be equal (Dijkhuizen et al., 1977). However, Park et al. (1986) demonstrated that preparations of SHb can contain both partially and completely sulphurated Hb tetramers. Tetramers containing sulphurated and unmodified subunits might show a reduced O₂ affinity because of the cooperative behaviour of Hb. In that case, the reduction in O₂ saturation would be greater than the percentage of SHb. These data indicate that, on the formation of SHb, the oxygen-binding capacity of the blood can be considerably reduced. In contrast to trout, however, the oxygen-binding capacity of carp red blood cells was not altered during sulphide exposure, which is in accordance with the negligible amount of SHb formed.

The present study demonstrates that significant amounts of SHb can be formed within the erythrocytes of rainbow trout. However, the 96 h LC₅₀ value for rainbow trout is only approximately 4 μmol l⁻¹ sulphide (Smith et al., 1976). The lowest sulphide concentration tested during this study was 100 μmol l⁻¹. Under these conditions, less than 0.5 % SHb was formed in trout haemolysate after 15 min of incubation.

Although SHb formation cannot play a role in acute sulphide toxicity in trout, sublethal levels of sulphide may induce significant formation of SHb when fish are exposed to low levels of sulphide for a long period. In contrast to trout, the common carp can survive concentrations of sulphide of more than 300 μmol l⁻¹ for several weeks (Kumar and Mukherjee, 1988). Blazka (1958) has shown that the muscle tissue of crucian carp (Carassius carassius) exposed to 1.7 mmol l⁻¹ sulphide contained approximately 1.2 mmol l⁻¹ sulphide after 8 h of exposure. Under these conditions, high sulphide concentrations can probably also be found in the blood. If we assume that a sulphide equilibrium will also be reached within a few hours in common carp exposed to 300 μmol l⁻¹ sulphide, blood sulphide concentrations of several hundred micromolar may occur under these conditions. SHb formation within red blood cells appeared to be very low even at extracellular concentrations of approximately 1 mmol l⁻¹. Nevertheless, SHb fractions may reach maximal values during long-term exposure to sulphide, and the carp’s ability to restrict SHb formation to only 10 % of total Hb may be an important contribution to the high sulphide-tolerance of this species.

Until now, it has been demonstrated that the Hb of three sulphide-tolerant fish species (common carp, this study; California killifish Fundulus parvipinnis and long-jawed mudsucker Gillichthys mirabilis, Bagarinao and Vetter, 1992) exhibit a low susceptibility to SHb formation. In contrast, the Hb of the sulphide-sensitive rainbow trout is much more susceptible to sulphide. Care must be taken, however, before concluding that a low rate of SHb formation represents a special adaptation to sulphide-rich environments. It will be necessary to study SHb formation in several more closely related species with different sulphide tolerance.

We are grateful to Norbert Heisler for reading and commenting on the manuscript. We would also like to thank Hannelore Schöder and Gabriele Meyer for skilful technical assistance and Arnold Stern for designing the software for spectral scans.