Respiratory adaptations in a deep-sea orbiniid polychaete from Gulf of Mexico brine pool NR-1: metabolic rates and hemoglobin structure/function relationships

Metazoan communities thriving on local chemolithoautotrophic and methanotrophic primary production were discovered in 1984 by the Geochemical and Environmental Research Group (GERG) on the Louisiana Slope of the Gulf of Mexico (Kennicutt et al.,1985; Brooks et al.,1987). Extensive beds of the methanotrophic symbiont-containing mussel Bathymodiolus childressi occur at many sites in the area(Childress et al., 1986;MacDonald et al., 1990a, b; Nix et al., 1995). Methanoaricia dendrobranchiata Blake (Polychaeta; Orbiniidae) occurs with the mussels at most of the sites and is particularity abundant in the mussel bed around Brine Pool NR1(Macdonald et al., 1990c),where it occurs either individually or in tight aggregations of as many as 150 worms (Smith et al., 2000; Hourdez et al., 2001). The worms often extend the anterior part of their body above the mussels into the overlying water (Fisher, 1996; Hourdez et al., 2001).

At Brine Pool NR1, the mussel bed rings the edge of the hypersaline pool on unconsolidated sediments saturated with brine. The oxygen concentration in water sampled between the mussels averaged 39 μmol l^-1 and was often undetectable by gas chromatography (<5 μmol l^-1; Smith et al., 2000). Although no sulfide could be detected in the brine pool itself, millimolar levels are present in the brine-saturated sediments under the outer edge of the mussel bed (Smith et al., 2000). The combination of sulfidic and hypoxic conditions presents a major respiratory challenge for active metazoans (Somero et al., 1989; Grieshaber and Völkel, 1998). Similar conditions characterize the microhabitats of many hydrothermal vent communities(Johnson et al., 1988; Fisher et al., 1988a, b; Sarrazin and Juniper,1999).

Several studies on hydrothermal vent polychaetes have shown larger mass-specific gill surface areas and shorter diffusion distances between the environment and the body fluids than in related genera from other environments(Jouin and Gaill, 1990; Hourdez and Jouin-Toulmond,1998). The gills of M. dendrobranchiata are also hypertrophied compared with those of other orbiniids and resemble those of the vent polychaetes (Hourdez et al.,2001). This characteristic and the short diffusion distance across the gill epithelia have been interpreted as adaptations for life in a hypoxic environment (Jouin and Gaill,1990; Jouin-Toulmond et al.,1996; Hourdez and Jouin-Toulmond, 1998; Hourdez et al., 2001).

In addition to these anatomical adaptations, the respiratory pigments of the vent polychaetes and vestimentiferan species studied to date have very high oxygen affinities and strong Bohr effects that may facilitate oxygen uploading in the gills at low oxygen tensions as well as its release in the tissues (Arp et al., 1990; Toulmond et al., 1990; Hourdez et al., 1999). M. dendrobranchiata contains an extracellular hemoglobin in its blood vessels (Hourdez et al.,2001). In annelids, such extracellular hemoglobins are usually giant hexagonal bilayer (HBL) molecules, made up of 12 large subunits(dodecamers) composed of globin chains. These dodecamers are held together by polypeptide chains known as linkers (for a review, see Lamy et al., 1996).

We studied aspects of the respiratory physiology of M. dendrobranchiata from the cold seeps of the Gulf of Mexico. The relationship between oxygen consumption rate and environmental oxygen concentration was determined together with tolerance of anoxia in both the presence and absence of sulfide. We also determined the concentration of hemoglobin in the worms and describe the molecular and functional properties of the purified molecule. Finally, these findings are discussed in the context of the animal's habitat and behavior.

Specimens of Methanoaricia dendrobranchiata Blake were collected together with mussels at the Brine Pool NR-1 site on the Louisiana Slope in the Gulf of Mexico (23°43′24″N, 91°16′30″W at 650 m depth; MacDonald et al.,1990c; Smith et al.,2000) in July 1998 using the scoop collector on the arm of the submersible Johnson Sea-Link I. The animals were placed in a temperature-insulated box on the submarine for recovery by the RV Edwin Link, where they were transferred to aquaria of chilled (8 °C)aerated sea water for between 1 and 7 days before use in oxygen consumption and anoxia experiments. There were no apparent adverse effects of maintenance at ambient pressure or normoxia on M. dendrobranchiata, which have been maintained alive in the laboratory, with mussels, for over a year.

The oxygen consumption rates of individual worms were determined in water-jacketed, gas-tight chambers (Fisher et al., 1985) at 8.5±0.5°C using Clark-type oxygen probes (Orion 835 dissolved oxygen meter)(Fig. 1). Oxygen concentrations in the chamber were recorded every 5 min for 16-17 h. After the experiment,the worms were rapidly dried with paper tissues and frozen in cryovials for transport to Pennsylvania State University. To determine the wet mass of the worm, the cryovial was weighed before and after removing the worm,Mass-specific oxygen consumption rates were calculated for each 20-min interval (using five data points) and plotted against the mean oxygen concentrations for that time interval.

Worms were maintained anaerobically in water containing 0, 0.06 or 1 mmol l^-1 sulfide to determine their tolerance to anoxia under these conditions. Oxygen-free sea water was obtained by bubbling with nitrogen for 30 min and then transferred to a nitrogen atmosphere in a glove bag in a cold room at 8 °C, where all subsequent manipulations were conducted. The absence of detectable oxygen was confirmed with a dissolved oxygen meter(Orion 835). Preweighed washed crystals of sodium sulfide were then added, and the sulfide concentrations were measured by gas chromatography(Childress et al., 1984). Individual worms were incubated separately in sealed 30 ml glass scintillation vials secondarily contained in a 500 ml gas-tight container filled with nitrogen-equilibrated water (to dampen minor temperature fluctuations and maintain anaerobic conditions). Eleven vials containing one worm each were incubated anaerobically under each experimental condition, and an additional 22 worms from the same collection were kept under normoxia in individual vials that were open to the atmosphere (controls). Survival was checked every 12 h until the ship reached port (7 days).

Because of the very small blood vessels, it was not possible to sample pure blood intravascularly. Approximately 75 worms from a single collection were batch-frozen 1 h after recovery for shipment to the laboratory and hemoglobin extraction. In the laboratory, the worms were homogenized with 100ml of extraction buffer (50 mmoll^-1 Tris, pH8, 1 mmoll^-1 EDTA and 1 μmoll^-1 phenylmethanesulfonic fluoride, PMSF). The homogenate was centrifuged for 20 min at 20 000 g to remove the coarse debris. The bright red supernatant was then centrifuged at 100 000 g for 5 h to pellet the hemoglobin. The pellet containing the hemoglobin was resuspended in 1 ml of the extraction buffer, and the hemoglobin was purified by fast protein liquid chromatography (FPLC) on a Superose 6 column (10 mm×300 mm; Pharmacia Inc., 5-5000 kDa separation range) using 400 mmoll^-1 NaCl, 2.95 mmoll^-1, KCl, 32 mmoll^-1 MgSO₄, 11 mmoll^-1 CaCl₂and 50 mmoll^-1 Hepes at pH 7.0 as the elution buffer. The elution rate was typically 0.5 ml min^-1, and the elutant was monitored at 280 nm. The molecular mass of the collected fractions was estimated using a sample of Riftia pachyptila blood, which contains four peaks of known molecular mass (Zal et al.,1996). The presence of hemoglobin in colored fractions was verified on the basis of peak absorptions near 414 nm.

To assess the significance of the hemoglobin for oxygen storage, 61 worms ranging widely in size (0.08-1.95 g) were individually frozen on board ship. In the laboratory, they were weighed and homogenized individually, using a ground-glass tissue homogenizer, in 1 mmoll^-1 Tris buffer, pH 8,containing 1 mmoll^-1 EDTA and 1 μmoll^-1 PMSF to prevent hydrolysis by proteases. The homogenate was centrifuged at 10 000 g for 15 min to remove cellular debris, and the supernatant was used for heme concentration measurements after appropriate dilution of the samples in the extraction buffer. Potassium ferricyanide solution (10 mmoll^-1; 10 ml) was added to 2 ml of this diluted solution, which had been incubated for 5 min at room temperature to oxidize the hemoglobin completely to methemoglobin. Potassium cyanide (50 mmoll^-1; 10 ml)was then added to obtain the cyan-methemoglobin derivative, which was quantified on the basis of the absorbance of the solution at 540 nm and the most appropriate published extinction coefficient of 11.0l mmol^-1cm^-1 (Van Assendelft,1970).

Images of negatively stained M. dendrobranchiata hemoglobin molecules were recorded by transmission electron microscopy to assess their molecular organization following a method modified from Valentine et al.(1968). Diluted purified hemoglobin (3 μl; 0.037 mmoll^-1 heme in 100 mmoll^-1Tris buffer at pH 6.85) was applied to a very thin (10-30 nm) carbon foil supported on a standard 3.05 mm copper electron microscopy grid. After allowing 1 min for the hemoglobin molecules to adhere to the carbon foil, 6μl of 2 % (m/v) uranyl acetate solution was applied to stain the molecules. After 10 s, the grid with the stained sample was blotted with Whatman filter paper, leaving only a very thin wet layer on the carbon foil, which was allowed to dry in air. The negatively stained specimens were examined using a Philips CM120 electron microscope at 100 keV, and images were recorded under minimum-dose conditions (Kenney et al.,1995) at 40 000× magnification.

The transmission electron microscopy images of the negatively stained hemoglobin molecules were analyzed to reveal the symmetry of their macromolecular structure. The images were digitized, and individual hemoglobin molecules were selected and centered by reference free alignment using the program SPIDER (Frank et al.,1996). The rotational frequency analysis of SPIDER-processed images was performed using the Medical Research Council image-processing programs (Crowther et al.,1996), and the rotational power spectra of the images were calculated.

Electro-spray ionization mass spectrometry (ESI-MS) was used under denaturing conditions to determine the molecular masses of the chains and covalently bound subunits present in the native hemoglobin and in the 210 kDa fraction (see Results). Three different treatments were applied to this hemoglobin sample to analyze its structure, as described previously (Green et al., 1996, 1999): (i)carboxyamidomethylation (Cam) to determine the number of free cysteines, (ii)reduction with dithiothreitol (DTT) to determine the composition of dimeric species and the reduced masses of the chains, and (iii) reduction plus Cam(reduced/Cam), to determine the total number of cysteine residues in each chain. The DTT reduction was carried out at a concentration of 5 mmoll^-1 for 2 min at 20 °C. Briefly, samples were introduced at a rate of 5 μl min^-1 into the electrospray source of a Quattro LC mass spectrometer (Micromass UK Ltd). They were analyzed at a concentration of 0.25-0.5 mg ml^-1 in 1:1 (by volume) acetonitrile:water containing 0.2 % formic acid. The scan range was m/z 600-2500, and the cone voltage (counter electrode to skimmer voltage) ramp was from 30 V at m/z 600 to 100 V at m/z 2500 (m is mass; zis charge). Mass scale calibration employed the multiply charged series of ions from horse heart myoglobin (M_r 16 951.5 Da; Sigma,M-1882). Data were processed using the Maximum Entropy (MaxEnt)-based software supplied with the instruments. Molecular masses are based on the atomic masses of the elements: C=12.011, H=1.00794, N=14.00674, O=15.9994 and S=32.066(IUPAC).

Equilibrium curves were obtained with a diffusion chamber apparatus(Sick and Gersonde, 1969)modified as described previously (Weber,1981). Briefly, small (4 μl) samples of purified hemoglobin were equilibrated with pure (>99.998 %) N₂ and O₂ and mixtures of these gases and air prepared using Wösthoff pumps. The pH of the hemoglobin solutions was varied by adding 6 μl of 1 moll^-1Bis-Tris Propane buffer of a range of pH values to 100 μl aliquots of the sample, and water was added to bring the total volume to 120 μl. The pH was measured in duplicate with a blood gas analyzer (BMS2, Radiometer) on 50 μl subsamples. Oxygen equilibrium and pH measurements were carried out at 10, 20 and 30 °C (±0.1 °C).

Values of P₅₀ (the P_O2at which the hemoglobin is half-saturated with oxygen) and n₅₀ (cooperativity at P₅₀) were derived from linear regression Hill plots of log[Y/(1-Y)]=f(logP₀₂)for S (saturation values) of 30-70 %. The Bohr factor (Φ) was calculated as Φ=ΔlogP₅₀/ΔpH, and the apparent heat of oxygenation was calculated asΔ H^obs=2.303RlogP₅₀/[1/T₁)-(1/T₂)],where R is the gas constant and T₁ and T₂ are the absolute temperatures. The value thus obtained(ΔH^obs) comprises the intrinsic heat of oxygenation(ΔH^intr), the heat of solution of O₂(ΔH^sol, approximately 13 kJ mol^-1) and contributions from oxygenation-linked processes such as the heats of reaction with protons and other effectors. In the absence of oxygenation-linked binding of protons (when Φ=0) and other allosteric effectors,Δ H^intr=ΔH^obs-ΔH^sol(see Wyman et al., 1977).

Values reported as means ± standard deviation (S.D.) throughout this manuscript.

Fig. 2A shows the reduction in oxygen concentration from 4 to 0.1 mgl^-1 (partial pressure decreasing from 10 000 to 260 Pa) recorded in the incubation chamber for a worm 1.24 g in wet mass. The calculated oxygen consumption rates for this experiment are shown in Fig. 2B. The oxygen consumption rate is relatively constant(0.66±0.08 μl O₂ min^-1 g^-1 wetmass, N=75 points) for oxygen levels between normoxia and 0.5 mg l^-1 (P_O2=1300 Pa). Below the critical value of 0.5 mg l^-1(P_O2=1300 Pa), oxygen consumption rate decreases gradually and ceases at a concentration of 0.1 mg l^-1(P_O2=260 Pa). The small variations in the mean oxygen consumption rate at the beginning of the incubation are interpreted as periods of activity and rest.

Separate linear regressions for points below 0.4 mg O₂l^-1 and points above 1.5 mg O₂ l^-1 allowed us to estimate the critical pressure (the intercept of the two lines). For the worm in Fig. 2, the intercept gives a critical pressure P_c of 0.27 mg O₂l^-1 (700 Pa).

These measurements were made for seven worms ranging from 0.64 to 1.87 g in wet mass. For two of these experiments, the oxygen concentration did not fall below 1 mg l^-1. In these cases, we used the linear portion of the oxygen concentration as a function of time curve(Fig. 2A) to calculate the oxygen consumption rate. For the five other experiments, the shape of the curve was always the same, and the χ²-value supported the use of the above equation for each data set (P<0.01). The P_c values ranged from 0.25 to 0.38 mg O₂l^-1 in the five experiments(P_O2=650-990 Pa) and oxygen consumption ceased at 0.09-0.11 mg O₂l^-1(P_O2=230-280 Pa).

Fig. 4 shows the percentage survival as a function of time under normoxia, anoxia, anoxia + 60μmoll^-1 sulfide and anoxia + 1 mmoll^-1 sulfide. None of the normoxic (control) animals died during the experiment, while only a single worm survived for 7 days under anoxia. The mortality curves are sigmoidal. Curve-fitting yield TL₅₀ values (time to reach 50 % survival) of 99.1, 119.6 and 133.5h, respectively, for anoxia + 1 mmoll^-1 sulfide, anoxia + 60 μmoll^-1 sulfide and anoxia without sulfide. Analysis of covariance (ANCOVA) of the linearized data indicated that the effect of the treatment was highly significant, and pairwise comparisons confirmed that all TL₅₀ values were significantly different (P<0.01).

The extracellular hemoglobin was purified by FPLC. Absorbances at 280 nm(which characterize proteins) and 414 nm (the Soret band, which characterizes hemoglobins) show two hemoglobin fractions with masses of approximately 3.5×10⁶ Da and 210 kDa. The second fraction, which was very small (elution profile not shown), is referred to as `dodecamers' (D in the figures) and is considered to be a dissociation product of the first fraction of whole hemoglobin molecules (W-Hb) (for an example, see Zal et al., 1996). Support for this designation is presented below.

The transmission electron microscopy pictures of the whole hemoglobin molecules (Fig. 6A,B) showed a strong peak at the sixth rotational harmonic, indicating a sixfold rotation symmetry (Fig. 6C). This information was used to produce a rotationally Fourier-filtered image. The result is a typical top (sixfold axis) view of a hexagonal bilayer (HBL)(Fig. 6D).

Under denaturing conditions, ESI-MS of the native hemoglobin revealed three groups of peaks with molecular masses of approximately 16, 32 and 49 kDa. In the first group, four peaks can be distinguished: a1, a2, a3 and a4, with masses of 15 824.8, 15 835.1, 15 864.5 and 16 255.4±1.5 Da and relative intensities of 0.40:0.34:0.19:0.17, respectively (N=5)(Fig. 7A; Table 1). A single peak (Dm)occurred at 31 972.6±3.0 Da (N=5), while the third group consisted of four peaks (LD1-LD4) between 48 715.5 and 48 804.9 Da(Fig. 7, inset in A). Careful examination of the ESI spectra from several preparations of W-Hb analyzed under various conditions failed to reveal the presence of components that could correspond to trimers or tetramers of globin chains. ESI-MS on the 210 kDa fraction revealed peaks corresponding to a1-a4 and Dm only (no peaks of higher mass were seen).

Under mild reducing conditions (1 and 2 min in the presence of 5 mmol l^-1 DTT at 22 °C) (Fig. 7B), the masses of a1-a3 remained unchanged, while the mass of a4 decreased by 119.5 Da, suggesting that this chain is cysteinylated. Concomitantly, the intensity of the Dm peak decreased and the group of peaks around 49 kDa disappeared. In addition, several new peaks appeared: b, c and d at 15 373.4, 16 600.6 and 17 172.9 Da, respectively (±1.5 Da, N=6), and L1, L2, L3, L4 and L5 at 24 358.7, 24 379.0, 24 404.2, 26 279.3 and 27 687.1 Da, respectively (±3.0 Da, N=4). These results imply that Dm is a disulphide-bonded dimer composed of monomers b and c, since b+c-2H=31 972.0, which is the mass determined for Dm in the unreduced hemoglobin (Green et al.,1996). Furthermore, twice the masses of L1, L2 and L3 minus 2H are within experimental error of the masses of LD1, LD2 and LD4(Fig. 7B), implying that L1-L3 exist as disulfide-linked dimers in the native hemoglobin. L1-L5 have masses that are typical of linker chains. However, no peaks were observed in the native hemoglobin that could correspond to dimers of L4 and L5. Moreover,while L1-L3 remained present upon reduction, L4 and L5 disappeared, with the concomitant appearance of several relatively minor peaks in the range 18-23 kDa. Peak d cannot be assigned to a higher-mass component in the W-Hb. It is possible that it is one of the components in a dimer or higher-order multimer that was not detected in the denatured W-Hb. It could also correspond to a fragment of a linker, made more fragile after reduction.

After carboxyamidomethylation (Cam) with iodoacetamide, the masses of a1-a3 remained unchanged, indicating that they contain no free cysteine. However,the mass of Dm increased by 2 Cam units, implying that it contains two free cysteine residues. The total number of cysteine residues in each chain and the reduced masses of the chains were determined from the reduced/Cam hemoglobin,as shown in Table 1 and Fig. 7C. Strong support for LD1-LD4 being dimeric forms of L1-L3 is provided by the reduced/Cam results as follows. L1-L3 have masses typical of linker chains and appear to exist as dimers in the native hemoglobin. The peak at 48 715.5 Da is probably a dimer made of two L1 chains (2L1-22H=48 710.2 Da) and the peak at 48 755.1 Da could be a dimer made of two L2 chains, or one L1 chain and one L3 chain (2L2-22H=48 757.8 Da and L1+L3-22H=48 757.1 Da, respectively). The peak at 48 777.4 Da is probably a dimer of L2 and L3 (L2+L3-22H=48 780.9 Da) and, finally, the peak at 48 804.9 Da is probably a dimer of two L3 chains (2L3-22H=48 804.0 Da). All these calculated masses are well within the experimental error for the measured masses of the peaks.

In Fig. 8, the effects of pH on the affinity (P₅₀) and cooperativity(n₅₀) of the intact hemoglobin (W-Hb) and dodecamers (D)at 10, 20 and 30 °C are shown. W-Hb and D exhibit the same oxygen affinities at pH 7.6 for 10 °C (P₅₀=27.8 Pa), at pH 7.4 for 20 °C (P₅₀=82 Pa) and at pH 7.1 for 30 °C(P₅₀=271 Pa). At lower pH values, D shows progressively higher affinities than W-Hb, whereas the reverse applies at higher pH.

Both hemoglobins express Bohr effects in the range pH 6.7-7.5. The Bohr factors in W-Hb vary inversely with temperature Φ=-0.48, -0.44 and -0.35 at 10, 20 and 30 °C, respectively, and exceed those in D (Φ=-0.11,-0.13 and -0.11, respectively).

W-Hb and D also differ in cooperativity coefficients, n₅₀. In W-Hb, the coefficient is higher, with a maximum at pH 7.5, and appears to vary inversely with temperature(n₅₀=2.5 at 10 °C, n₅₀=2.0 at 20°C and n₅₀=1.8 at 30 °C). In D the value of n₅₀ (1.2) is almost independent of pH.

Extended Hill plots of oxygen equilibria of the intact hemoglobin molecules(W-Hb) at pH 6.77 and 7.60 are shown in Fig. 9. The allosteric and other parameters derived from analysis of the equilibria in terms of the two-state Monod—Wyman—Changeux (MWC) allosteric model (as described by Weber et al.,1995) are given in Table 2. The correspondence between the cooperativity value at half-saturation (n₅₀) and the maximum cooperativity(n_max) and between P₅₀ and the median oxygen affinity P_m(Table 2) reflects symmetrical oxygen-binding curves and permits the use of P₅₀ values to analyse allosteric interactions. The control mechanism of the Bohr effect in this hemoglobin is primarily an increase in the oxygen-binding affinity of the hemoglobin in the oxygenated R (`relaxed') state (K<sub>R</sub>) as pH rises,while the affinity (K<sub>T</sub>) of the deoxygenated T (`tense') state remains relatively constant (Table 2). This raises the free energy of cooperativity(▵G) (Table 2)and the K_T/K_R ratio with increasing pH. The analyses indicate the presence of 3-5 interacting oxygen-binding sites (Q) in the functional subunits of the intact hemoglobin molecules.

The oxygen affinities of both hemoglobins show strong temperature-dependence (Figs 8, 10) at pH 6.8 and 7.6. The temperature effect is smaller in W-Hb than in D(▵H^obs=-58 and -66 kJ mol^-1 for W-Hb at pH 6.8 and 7.6, respectively, and -77 kJ mol^-1 for D at both pH values).

In contrast to the marked effect of 100 mmol l^-1Ca²⁺, which increases the oxygen affinity of W-Hb(▵logP₅₀=0.28)(Fig. 11), D lacks sensitivity to Ca²⁺.

Methanoaricia dendrobranchiata is found in high densities at cold-seep sites often in association with mussels on the Upper Louisiana Slope(Blake, 2000; Smith et al., 2000). While this environment is characterized by a high local production (a strong trophic attractor), it provides harsh chemical conditions for the survival of metazoans (low oxygen and high sulfide concentrations). To survive and thrive under these conditions, the polychaetes must be adapted to take up oxygen at low environmental partial pressures and to tolerate anoxia and sulfide exposure.

Intact Methanoaricia dendrobranchiata hemoglobin has a native mass and appearance (as seen in transmission electron microscopy) typical of giant HBL annelid hemoglobins (for a review, see Lamy et al., 1996). On the basis of the masses of its dissociation products, it contains both globin and linker chains, as expected for HBL hemoglobins (for a review, see Lamy et al., 1996). There appear to be five linkers, although three of them (L1-L3) behave quite differently from the other two (L4 and L5) when reduced. Linkers L1-L3 possess 11 cysteine residues, 10 involved in intra-chain disulfide bonds and one involved in a disulfide bond between two linker chains(Table 1). All the linkers are dimeric, as in Arenicola marina(Zal et al., 1997), and are stable as monomers under reducing conditions. However, the other two linkers(L4 and L5), if indeed they are linkers, only appeared as monomers under mild reducing conditions and were never observed as either monomers or dimers in the native hemoglobin. Upon further reduction, they disappeared, and a number of components in the mass range 18-23 kDa were observed; these then disappeared after a longer reduction time. At this time, monomeric L1-L3 were still present, suggesting that L4 and L5 are significantly less stable than L1-L3. Both monomeric (a1-a4) and dimeric (b and c) globins are present in M. dendrobranchiata hemoglobin. The globin dimers were unexpected because dimers have previously been considered to be typical of hirudinean(leech) and vestimentiferan hemoglobins in contrast to all other studied polychaetes, which possess globin trimers (for a review, see Lamy et al., 1996). This is also the first record of an annelid globin found to be cysteinylated. The origin of d remains unknown: it is probably involved in higher aggregation states, not detected by ESI-MS, or corresponds to a fragment of a linker,probably L4 or L5. A similar situation, where some subunits do not seem to ionize properly under non-reducing conditions, was encountered in Eudistylia vancouverii chlorocruorin(Green et al., 1998).

Both intact molecules and dodecamers exhibit similar and very high oxygen affinities but differ in other oxygen-binding characteristics (cooperativity,Bohr effect, sensitivity to Ca²⁺ and sensitivity to temperature). Their affinities are among the highest reported in annelids and probably represent adaptations to the animal's low-oxygen environment(Toulmond et al., 1990; Weber, 1978a, 1980; Toulmond, 1992; Hourdez et al., 1999, 2000; Weber and Vinogradov, 2001). The increase in K_R with rising pH(Fig. 9) is opposite to the situation for human hemoglobin (in which the Bohr effect is due to variation in K_T; Tyuma et al.,1973), but similar to that in the polychaete Arenicola marina, where it increases the Bohr factor at high oxygen saturation and favors oxygen saturation of blood entering the relatively alkaline gills(Weber, 1980; Toulmond, 1992).

The Bohr effect in the intact molecules (W-Hb) is similar to that of most extracellular hemoglobins (Weber,1978a) but is greatly reduced in the dodecamer, as previously reported for other annelids (for a review, see Lamy et al., 1996).

In contrast to similar blood Ca²⁺ concentrations among vestimentiferans and other annelids studied (7-13 mmol l^-1), blood Mg²⁺ levels vary widely among species (from 2 to 44 mmol l^-1) (Oglesby,1978; Sanders and Childress,1991) and thus may affect the functional properties of annelid hemoglobins in vivo. Ca²⁺ increases the oxygen affinity of intact M. dendrobranchiata hemoglobin but has no effect on the dodecamers, suggesting that the effect of group IIA cations such as Ca²⁺ and Mg²⁺ involves the linkers. This agrees with other studies on HBL hemoglobin and most studies of dodecamers, with the exception of Lumbricus terrestris hemoglobin dodecamers, which are sensitive to group IIA cations (for a review, see Lamy et al., 1996). Lumbricus terrestris hemoglobin has been shown to have Ca²⁺-binding sites with a variety of affinities that mainly involve linkers (Kuchumov et al.,2000). The low-affinity sites are likely to be involved in the effects of group IIA cations on the structure and function of hemoglobin(Kuchumov et al., 2000).

The cooperativity coefficients of M. dendrobranchiata hemoglobin are within the range previously reported for HBL hemoglobins(Weber and Vinogradov, 2001). The value for the intact molecule (2.5) is approximately twice that of the dodecamer (1.2), as seen in Lumbricus terrestris hemoglobin(n≈5 and n≈2, respectively, at pH 7.2; Krebs et al., 1996). The difference suggests a participation of the linker chains in heterotropic interactions in the molecule, as in other HBL hemoglobins (for a review, see Lamy et al., 1996).

The temperature-sensitivity of the intact hemoglobin (▵H=-66 kJ mol^-1 at pH 7.6) is similar to that of the polychete Abarenicola clarapedii, another annelid that lives in a fairly stenothermal environment, but greater than that of Arenicola marina(▵H=-22 kJ mol^-1; Weber, 1972), which experiences larger temperature variations in its natural habitat. Given that dissociation of Bohr protons is endothermic and reduces the net heat liberated upon oxygenation of the hemes, the lower temperature effect in the whole molecules than in the dodecamers (Fig. 10) accords with the larger Bohr effect in the former. Analogously, ▵H^obs of the intact molecules is lower at pH 6.8 than at pH 7.6, where the Bohr effect is smaller (see Fig. 8).

The high affinity and strong Bohr effect observed in the vascular hemoglobin probably have adaptive significance in a low-oxygen environment,favoring both the binding of oxygen at the respiratory surfaces and its release in the tissues, where pH decreases as a result of the production of acid excretory products.

The oxygen consumption rates of M. dendrobranchiata are within the lower end of the normal range of respiratory rates reported for polychaetes(Childress and Mickel, 1985; Weber, 1978b). The worms are very good oxyregulators and are able to maintain a constant metabolic rate as environmental concentrations fall to 0.3 mg O₂ 1^-1(where Po₂=870 Pa)(Fig. 2). At lower partial pressures, the oxygen consumption rate drops sharply, and oxygen consumption ceases below 0.1 mg O₂ 1^-1 (Po₂=260 Pa).

Strong oxyregulation extending to very low oxygen concentrations is normally correlated with one or more of the following adaptations: (i) a large gas-exchange surface, (ii) a short diffusion distance between the environment and the body fluids to facilitate gas diffusion, (iii) the presence of a respiratory pigment to increase the oxygen-carrying capacity of the circulating body fluid(s), (iv) a high oxygen affinity of the respiratory pigments to increase oxygen loading at low oxygen tension, (v) a high ventilatory rate to renew the medium as it becomes depleted of oxygen, and(vi) a high circulatory rate. M. dendrobranchiata have a large gill surface area and reduced diffusion distances across their respiratory surface as a result of intra-epidermal blood vessels in the gills(Hourdez et al., 2001). Furthermore, the worms are capable of greatly extending their gills when exposed to hypoxic and anoxic conditions (S. H., personal observation). The presence of cilia on the surface of the gills(Hourdez et al., 2001)facilitates renewal of the diffusion-limited water layer. These worms do not ventilate their gills, and no information is available about ciliary movement under hypoxia. However, mussels from their natural habitat can create strong,small-scale currents.

M. dendrobranchiata hemoglobin has a very high oxygen affinity under physiological conditions; an oxygen partial pressure of 100 Pa, which is far below Pc, saturates the hemoglobin to 96 % at 10 °C in vitro. However, the oxygen consumption rate of M. dendrobranchiata decreases when the oxygen partial pressure falls below 870 Pa, probably as a result of diffusion-limitation driven by the decreased oxygen concentration gradient across their gills. However, this value (870 Pa) is low, indicating that the high oxygen affinity partially offsets the lack of ventilatory capability. The low P_c value reflects a high degree of oxyregulation,which is similar to that in Riftia pachyptila but slightly greater than those in the crab Bythograea thermydron and the clam Calyptogena magnifica, all from deep-sea hydrothermal-vent habitats(Childress and Fisher, 1992). Many deep-sea hydrothermal-vent and cold-seep microhabitats are characterized by low levels of dissolved oxygen, and endemic animals often show similar capacities to oxyregulate, delaying the onset of anaerobic metabolism.

Although the high mass-specific gill surface area favors oxygen uptake, it cannot selectively favor the entry of oxygen without permitting the inward diffusion of potentially harmful sulfide(Somero et al., 1989), which often occurs at substantial concentrations in the mussel beds(Smith et al., 2000; Nix et al., 1995). Sulfide further compromises respiration in most animals by inhibiting respiration at the level of cytochrome c oxidase and hinders binding of oxygen to most hemoglobins (Somero et al.,1989). The sulfide-tolerance under anoxia of these worms compares with that of Marenzelleria cf. wireni and is the highest found among the species studied to date(Fig. 12; Groenendaal, 1980; Schiedeck et al., 1997; Gamenick et al., 1998). The hemoglobin-bound oxygen store is not enough to sustain aerobic metabolism for more than 31 min. However, worms faced with prolonged anoxia switch to anaerobiosis and can survive for days even in the presence of sulfide.

Cold seeps, like hydrothermal vents, are habitats characterized by very high productivity and biomass compared with the generally nutrient-poor deep sea. The high level of endemism and low species-richness of these communities suggest that these environments are also physiologically challenging and that they require specific adaptations. We suggest that the ability of Methanoaricia dendrobranchiata to colonize their extreme environment is due at least in part to the adaptations reported here, such as their capacity to oxyregulate, which is due to the presence of a high-affinity hemoglobin, and their tolerance to anoxia and sulfide. These adaptations allow the worm to forage in the sulfidic, organic-rich sediments between and below the mussels and to exploit a habitat with little or no competition and limited predation (MacAvoy et al.,2002).

This work was supported primarily by subcontract L100094 to the Mineral Management Service project RF-6899 and the Minerals Management Service, Gulf of Mexico Regional OCS Office, through contract number 1435-01-96-CT30813, and the Danish Natural Science Research Council. We thank Anny Bang, Århus,Denmark, for technical assistance and Catherine Venien-Bryan, Laboratory of Molecular Biophysics, Oxford, UK, for the use of the electron microscope facilities. This work would not have been possible without the expert assistance of the captain and crew of R/V Edwin Link and the pilot and crew of the DSRV Johnson Sea-Link.