Hemoglobin function in deep-sea and hydrothermal-vent endemic fish: Symenchelis parasitica (Anguillidae) and Thermarces cerberus(Zoarcidae)

Deep-sea hydrothermal vents provide some of the harshest aquatic conditions that metazoans are exposed to in nature. Here, hot (350°C), anoxic waters bearing high levels of H₂S (4-8 mmol l^-1) heavy metals,arsenic compounds and CO mix chaotically with cold (2°C) deep-sea,O₂-laden (0.11 mmol l^-1) water at extremely high pressures (∼25.3 MPa at 2500 mdepth; Von Damm, 1990, 1995; Magemheim and Gieskes, 2002). Hemoglobins (Hbs) from invertebrates colonizing these habitats display striking adaptations, including high affinities for oxygen and pronounced Bohr effects, compared with those living in well-oxygenated environments(Terwilliger and Terwilliger,1984; Arp et al.,1990; Toulmond et al.,1990; Hourdez et al., 1999, 2000a,b)and additionally may participate in the detoxification of sulfide (cf. Weber and Vinogradov, 2001). A well-documented example is the Hb of the giant vestimentiferan tube-worm Riftia pachyptila, which binds sulfide reversibly with a high affinity and transports the reduced compound to symbiotic chemoautotrophic bacteria that oxidize sulfide and fix CO₂ inside the body(Arp et al., 1987).

Fish show pronounced, well-documented adaptations in blood O₂-binding characteristics in response to environmental stresses(Weber, 1996) that comprise intraspecific adaptations (such as the decreases in the red cell levels of organic phosphates, ATP and guanosine triphosphate (GTP), that increase Hb–O₂ affinity and occur in individual specimens) as well as interspecific adaptations (between-species differences that are genetically coded and commonly involve differences in Hb structure and heterogeneity).

Fishes show marked isoHb differentiation and may be categorized accordingly. Whereas `class I' species, such as cyprinids and ciclids(Gillen and Riggs, 1972),possess only electrophoretically anodic Hb components with relatively low O<sub>2</sub> affinities and pronounced Bohr and Root effects (decreases in O<sub>2</sub> affinity and carrying capacity, respectively, at low pH that promote O<sub>2</sub> unloading in the respiring tissues and in the swimbladder), `class II' fish (including eels, salmonids and some catfish)additionally have cathodic Hbs that exhibit high intrinsic O₂affinities and low pH sensitivities(Powers and Edmundson, 1972; Gillen and Riggs, 1973; Weber et al., 1976a, 2000; Pellegrini et al., 1995) and may function as a reserve transport system when oxygenation of the anodic components is compromised by acidification and hypoxia (Weber, 1990, 2000a).

With no information available on Hb function in hydrothermal-vent vertebrates, we investigated isoHb differentiation, together with Hb–O₂ binding and its sensitivity to pH, temperature and red cell effectors (organic phosphate and chloride ions), in the anguillid pugnose eel Symenchelis parasitica and the zoarcid Thermaces cerberus and carried out some comparative measurements on the Hb of the temperate, shallow-water zoarcid Zoarces viviparous. Compared with S. parasitica, which shows highly cosmopolitan distribution at depths of up to 3000 m and lives at the borders of hydrothermal vents that it casually visits for food(Desbruyères and Segonzac,1997), T. cerberus is endemic to hydrothermal-vent environments, where it intertwines with, and feeds on, Riftia(Dahlhoff et al., 1990; Geistdoerfer and Seuront,1995), which requires sufficient quantities of free sulfide to support robust bacterial endosymbiosis(Luther et al., 2001).

The pugnose eel Symenchelis parasitica (Goode and Bean 1879)(Synaphobrachidae, Anguilliform) was captured using fish traps at a depth of 1690 m at the `Lucky Strike' vent field of the Mid-Atlantic Ridge (MAR)hydrothermal vents (37°17′ N and 32°16′ W) during the MARVEL'97 expedition. S. parasitica is a benthopelagic, necrophagus fish encountered on continental slopes and at upper abyssal ocean depths and was previously believed to be parasitic due to its sucker-shaped mouth. Specimens of the vent-endemic eelpout Thermarces cerberus (Rosenblatt and Cohen 1986) (Gadiformes, Zoarcoidei, Zoarcidae) were captured in Riftia clumps at 2500-2650 m depth at the East Pacific Rise(12°49′ N, 103°56′ W and 9°46′ N,104°21′ W) during the HOT'96 cruise. The deep-sea fish used were approximately 30 cm long. The north-European eelpout Zoarces viviparus Linnaeus 1758, which was used for comparative measurements,originated from Aarhus Bay, Denmark.

Blood samples from the deep-sea species were drawn into EDTA-containing syringes, frozen and stored at -80°C until use. Hb solutions were prepared by addition of approximately two volumes of 0.02 mol l^-1 Tris buffer, pH 7.6 and stripped of organic phosphates on MB-1 mixed ion-exchanger or by preparative isoelectric focusing. Other preparatory steps were carried out at 0-5°C, as previously described (Weber et al., 1987, 2000). IsoHb composition was investigated by electrophoresis on cellulose acetate strips (Gelman Science,Ann Arbor, MI, USA) and isoelectric focusing in 110 ml(LKB) columns containing Pharmacia (Biotech AB, Uppsala, Sweden) ampholines, pH 3.5-10 (0.55%) and 5-8(0.18%). pH values of retrieved fractions were measured at 22°C using a BMS Mk2 Blood Micro System (Radiometer, Copenhagen, Denmark). Z. viviparus Hb was prepared from washed red cells and stripped by filtration through a column of Sephadex G25 (fine) gel. All Hb solutions were dialysed against three changes of 0.01 mol l^-1 Hepes buffer, pH 7.67 containing 0.5 mmol l^-1 EDTA and, where necessary,concentrated by ultrafiltration (Millipore 10 000 NMWL Ultra-free-4 filters). The Hb was frozen in 50 μl or 100 μl aliquots that were freshly thawed for O₂ equilibrium measurements.

Oxygenation equilibria of ultrathin (≤0.05 mm) layers of Hb solutions were recorded using a modified O₂-diffusion chamber(Weber, 1981; Weber et al., 1987) in the absence or presence of ATP (assayed using Sigma test chemicals) and 0.1 mol l^-1 KCl (assayed using a Radiometer CMT10 chloride titrator). The pH was varied using Hepes buffers (final concentration 0.1 mol l^-1; Weber, 1992). The overall heat of oxygenation {ΔH′ = R log_e[ΔlogP₅₀/(T₁^-1–T₂^-1)],where P₅₀ is the half-saturation O₂ tension, R is the gas constant, and T₁ and T₂ are different absolute temperatures; Wyman,1964} was investigated by measuring P₅₀ values at 5°C, 25°C and 35°C and at pH values near 6.8 and 7.5 and interpolating the P₅₀ at these two pH values from the linear logP₅₀vs pH regressions.

Isoelectric focusing (Fig. 1A) resolved S. parasitica Hb into a major component with an isoelectric point (pI) of 8.0 (Hb II), a minor cathodic component, Hb I (pI∼9.0), and two small anodic components, Hb III (pI ∼6.5) and Hb IV (pI∼5.2, not shown). Planimetric analysis indicates relative proportions of approximately 13:82:5 for Hbs I, II and III, respectively. Mindful of the differentiation between cathodic (pI>8.6) and anodic (pI<8.6) fish Hbs,functional analyses focused on Hbs I and II. T. cerberus Hb resolved into three major anodic components (Hbs I, II and III) with pI values of approximately 7.0, 6.8 and 6.5, respectively, and five minor components with pIs of <6.5 and >7.0 (Fig. 1B).

S. parasitica Hbs I and II exhibit radically different O₂-binding properties. Stripped Hb I has a markedly higher affinity than Hb II [P₅₀ at pH 7.2 = 14 Torr (1.87 kPa)and 33 Torr(4.40 kPa), respectively, at 25°C] and higher cooperativity at half-saturation (n₅₀=1.7 and 1.1, respectively; Figs 2, 3).

Hb I shows a slight, reverse Bohr effect(φ=ΔlogP₅₀/ΔpH=0.08) that is unaffected by chloride but changes to a slight, normal Bohr effect (φ=-0.14) in the presence of Cl^- + ATP (Fig. 3A). By contrast, stripped Hb II has a pronounced Bohr effect(φ=-0.39) that is potentiated by both Cl^- (φ=-0.58) and Cl^- + ATP (φ=-0.74) in accordance with increased anionic binding at low pH. Hb II shows a reverse `acid' Bohr effect below pH 6.7(Fig. 3A), which thus extends to higher pH values than those occurring below pH 6.0 in anodic Hbs of trout,carp and catfish Hbs (Binotti et al.,1971; Gillen and Riggs,1977; Weber et al.,2000). The P₅₀, n₅₀ andφ values of the stripped hemolysate(Fig. 3A) are intermediate between those of Hb I and Hb II, indicating an absence of interaction between the components.

In contrast to the moderate Cl^- and ATP sensitivity of Hb II, Hb I lacks Cl^- sensitivity (Figs 2, 3) but exhibits strong Cl^- + ATP sensitivity. The persistence of a pronounced ATP effect in Hb I at high pH (Fig. 2A) is intuitively consistent with its high pI value, reflecting positively charged sites. The induction of a normal Bohr effect of Hb I by ATP reflects preferential phosphate binding at low pH and thus is analogous to the increased normal Bohr effect in Hb II in the presence of ATP. The data at 5°C (Fig. 3B) confirm the functional differentiation between Hb I and Hb II but show a larger Cl^- + ATP effect than at 25°C(ΔlogP₅₀ at pH 7.5 = ∼0.6 and ∼0.4,respectively). Curiously, a distinct reverse Bohr effect (confirmed repeatedly) is manifest in Hb II at 5°C in the presence of Cl^-+ ATP, although a large (normal) Bohr effect is seen in the absence of the phosphate (Fig. 3B). With no information on the structure of the Hbs, a molecular explanation cannot be offered.

Given that cytoplasmic domains of the red cell membrane protein Band 3(cd-B3) bind at the phosphate-binding site of human deoxyHb(Walder et al., 1984), we investigated the effect of trout cd-B3, a synthetic 10-mer peptide corresponding to the amino terminus of trout Band 3 protein(Jensen et al., 1998), on S. parasitica Hb but found no detectable effect over a wide range of pH conditions (6.6-7.5; Fig. 3A).

Extended Hill plots (Fig. 4)of the major S. parasitica isoHb (Hb II) indicate negative cooperativity at extreme, low O₂ tensions and positive cooperativity only above ∼60% O₂ saturation, indicating that the molecules remain frozen in the deoxygenated `tense' conformation at low O₂ saturations. As is evident from Fig. 4, increased temperature as well as decreased pH lower O₂ affinity of Hb II predominantly by decreasing K_T without significantly impacting K_R (the O₂ association constants of the low-affinity deoxy and the high-affinity oxy states of the molecules,respectively), revealing greater Bohr effect and ΔH′values in the deoxygenated compared with the oxygenated state. Curiously, at low pH (7.0) and 5°C, Hb II shows distinct negative cooperativity(n<1) at 30-50% O₂ saturation in the absence of ATP(Fig. 4), which correlates with the high P₅₀ value (and Bohr factor) found under these conditions (Fig. 3B).

T. cerberus Hb exhibits a strikingly higher O₂ affinity than does Z. viviparous hemolysate [P₅₀=9 Torr(1.2 kPa) and 30 Torr (4.0 kPa), respectively, at pH 7.0] and a lower pH Bohr factor (Fig. 5), whereby the affinity difference between the two species increases with decreasing pH. The Bohr factor decreased with increasing temperature (φ at pH 7.0-7.5=-0.62,-0.56 and -0.37 at 15°C, 25°C and 35°C, respectively), indicating temperature dependence of ionization groups as reported in other vertebrate Hbs (Antonini and Brunori,1971), and falls drastically at high pH (>8; Fig. 5A). In contrast to the marked functional differentiation between S. parasitica isoHbs, T. cerberus isoHbs (I, II and III) show similar O₂affinities [P₅₀=6-8 Torr (0.8-1.07 kPa) at pH 7.0 and 25°C], similar low chloride sensitivities, similar pronounced ATP effects(Fig. 6) and similar cooperativities (n₅₀=∼1.5).

The O₂ affinities of S. parasitica and T. cerberus Hbs at 5°C, 25°C and 33-35°C yield essentially linear van't Hoff plots (Fig. 7), indicating temperature independence of the oxygenation enthalpies (ΔH′), and similar heat capacities in the oxygenated and deoxygenated states of the Hb (cf. Fago et al., 1997a). The oxygenation enthalpies comprise the exothermic intrinsic heat of heme oxygenation (ΔH⁰), the heat of solution of oxygen(ΔH^sol=∼-13 kJ mol^-1) and endothermic contributions that include the oxygenation-linked dissociation of hydrogen ions, organic phosphate or chloride ions (ΔH^H,Δ H^P and ΔH^Cl,respectively). It follows that the Bohr effect and phosphate binding reduceΔ H′ and that ΔH⁰ can experimentally be assessed in the absence of oxygenation-linked ion binding under pH conditions where there is no Bohr effect.

S. parasitica Hb II shows high temperature sensitivity(ΔH′=-52 kJ mol^-1) at pH 8.0, where there is almost no Bohr effect, and a lower sensitivity (-31 kJ mol^-1) at pH 6.8, where the Bohr factor is pronounced, suggesting that the intrinsic value(in the absence of proton and anion binding) exceeds -52 kJ mol^-1. As may be interpolated from Fig. 4, the heat of oxygenation is higher in the deoxygenated state than at half saturation (ΔH_T=∼-70 kJ mol^-1 and ∼-40 kJ mol^-1, respectively, at pH 7.0). The cathodic Hb I, which has a slight, reverse Bohr effect (cooperativity between proton and O₂ binding), shows lower enthalpies(ΔH′=-46 kJ mol^-1 and -49 kJ mol^-1,respectively, at pH 6.8 and 7.5).

The temperature sensitivities of Hbs from the vent-endemic and temperate zoarcids T. cerberus and Z. viviparus show corresponding patterns (Fig. 7B) despite the large difference in O₂ affinities. At pH 8.4, where these Hbs only show slight Bohr effects (normal and reverse, respectively; Fig. 5A,B), the respective enthalpies of -72 kJ mol^-1 and -78 kJ mol^-1 suggest that the intrinsic heat for the reaction with O₂ is intermediate(∼75 kJ mol^-1). The lesser reductions seen in T. cerberus Hbs compared with Z. viviparus Hbs under physiological conditions (ΔH′=-63 kJ mol^-1 compared with -42 kJ mol^-1, respectively, at pH 7.5, and -48 kJ mol^-1 and-31 kJ mol^-1, respectively, at pH 6.8; Fig. 7B) tally neatly with the smaller Bohr factors in the former species.

Extreme environmental conditions mandate integrated adaptations at systemic, cellular and molecular levels of biological organisation. Environmental factors that challenge Hb–O₂ transport in animals living at or near hydrothermal vents include hypoxia and low pH, which reduce O₂ binding, and high CO and H₂S levels, which potentially block O₂ transport and aerobic metabolism. Deep-water fish species exposed to high pressure generally exhibit lower oxygen consumption rates and enzyme activities and higher membrane fluidities than do shallow-water fish species (Sebert,2002). The hematological characteristics of deep-sea fishes similarly reflect adaptations to low metabolic and activity rates: Lycodes esmarkii, Antimora rostrata and Macrurus berglax from a depth of 280-1400 m have low blood Hb levels, large erythrocytes and hyperbolic blood O₂ curves (Graham et al.,1985).

A relevant consideration pertaining to deep-sea habitats is how high hydrostatic pressure influences water O₂ tension and the O₂ affinity of blood and Hb. Thermodynamic considerations indicate that the solubility of O₂ decreases whereas its partial pressure increases with increasing pressure and depth(Fenn, 1972). Increasing hydrostatic pressure to 1000 atmos (101.3 MPa) raises the affinity of human and menhaden (Brevoortia tyrannus) Hb approximately 2-fold without disturbing the transition between the deoxygenated, low-affinity (T) state and the oxygenated, high-affinity (R) states of the Hb(Carey et al., 1977). Increasing pressure from 1 atmos (0.1 MPa) to 126 atmos (12.8 MPa) raises O₂ affinity of human whole blood, red cell suspensions and hemolysate without affecting the sensitivity to 2,3-diphosphoglycerate (DPG; Reeves and Morin, 1986). Similar affinity increases have been reported in most earlier studies,although these often suffer from pitfalls associated with the use of buffers with pressure-sensitive pK values and the pressure sensitivities of gas solubility and spectral absorbances(Reeves and Morin, 1986).

Distinct adaptations to ambient conditions are seen. Thus Hb–O₂ affinities are much higher in vent-endemic T. cerberus, which grazes amongst sulfide-metabolising Riftia, than in S. parasitica, which frequents cold, O₂-laden deep-sea water and only casually visits the organic-rich vent areas. Also, whereas Hb–O₂ affinities in T. cerberus are higher than in Z. viviparous, those in S. parasitica are lower than in the eel Anguilla anguilla (Fig. 8). However, in contrast to the striking adaptations encountered amongst Hbs of hydrothermal-vent invertebrates, the Thermarces and Symenchelis Hb systems exhibit the same basic functional differentiations as encountered in shallow-water class I and class II fish,respectively. This aligns with the view that the regulatory burden for environmental adaptations in vertebrates is shifted to higher (e.g. cellular and organismic) levels of organisation than in invertebrates(Weber and Vinogradov,2001).

In humans, glycolytic enzymes compete with deoxyHb for binding cd-B3,providing a mechanism whereby Hb oxygenation can govern red cell glycolytic processes (Giardina et al.,1995; Messana et al.,1996). The insensitivity of S. parasitica Hb to trout cd-B3 (Fig. 3A) suggests that S. parasitica Hb does not play a transducer role. This agrees with data for salmonid (trout) isoHbs (Jensen et al., 1998; Weber,2000a) but contrasts with human and catfish (Hoplosternum littorale) Hbs (Walder et al.,1984; Weber et al.,2000; Weber,2000a), whose O₂ affinities are lowered by cd-B3,possibly implicating these Hbs in regulating cellular metabolism in an oxygenation-dependent manner (Weber,2000b).

The Hb systems of T. cerberus and S. parasitica appear to be typical representatives of fish classes I and II, respectively(Fig. 8).

IsoHbs from the vent-endemic eelpout T. cerberus have similar O₂ affinities and pronounced, normal Bohr effects that decrease with increasing pH, where the proton-binding sites become neutralized. A significant adaptation in T. cerberus Hbs appears to be the high O₂ affinity compared with that in the temperate eelpout Z. viviparus [P₅₀=∼8 Torr (1.07 kPa) and ∼24 Torr (3.2 kPa), respectively, at pH 7.0 and 25°C]. The affinity difference will be further amplified in the presence of the natural compliment of red cell effectors [P₅₀ values become ∼20 Torr (2.67 kPa)and ∼60 Torr (8 kPa) in the presence of ATP and 0.1 mol l^-1Cl^-; Fig. 8]. Although the exact mechanism of high affinity in T. cerberus must await the solution of their molecular structures, reduced sensitivities to chloride and phosphate effectors appear not to be involved(Fig. 8).

S. parasitica typifies class II fish, having cathodic Hb with relatively high O₂ affinity, a slight, reversed Bohr effect in the absence of organic phosphate and a large effect of ATP (that normalizes the Bohr effect) as well as an anodic Hb with relatively low affinity and marked Bohr and ATP effects – as found in other anguillids(Weber et al., 1976a; Fago et al., 1995; Tamburrini et al., 2001) and catfish (Garlick et al., 1979; Powers and Edmundson, 1972; Weber et al., 2000; Fig. 8). Unexpectedly, the affinities of stripped S. parasitica Hbs [P₅₀=14 Torr (1.89 kPa) and 33 Torr (4.4 kPa) for Hb I and II, respectively, at pH 7.2 and 25°C] are low compared with those obtained by the same technique in eel and catfish cathodic and anodic Hbs [P₅₀=≈2 Torr(0.23 kPa) and ≈8.5 Torr (1.13 kPa), respectively; Weber et al., 1976a; Fago et al., 1995, 1997b; Weber, 2000a] but are similar to those in trout Hbs [P₅₀=∼17-20 Torr (2.27-2.67 kPa)at 20°C; Weber et al.,1976b] – all of these species are classified as class II.

The pH insensitivity of the cathodic Hbs favors O₂ binding under acidotic conditions (burst activity, acid influx or lactate secretion in the swimbladder; Powers, 1972; Weber, 1990). In S. parasitica, however, the division of labor between the cathodic Hb I and anodic Hb II may be of limited physiological significance due to the low abundance of Hb I (cf. Fig. 1). The higher sensitivity to ATP in cathodic Hb I compared with Hb II aligns with Hbs of the eel Anguilla (Weber et al., 1976a; Fago et al.,1995) and the Amazon fishes Myllossoma, Pterygoplicthysand Hoplosternum (Martin et al.,1979; Weber and Wood,1979; Weber et al.,2000) but contrasts with rainbow trout (Oncorhynchus mykiss), whose cathodic Hb I is insensitive to phosphates. This confirms that the `model' trout Hb system is exceptional rather than prototypical. An intermediate situation appears in the South African mudfish Labeo,where the phosphate sensitivity of cathodic Hb I is markedly lower than those of the anodic isoHbs (Frey et al.,1998). In cathodic eel and Hoplosternum Hbs, theβ-chain amino acid residues implicated in phosphate binding are Val1(NA1), Glu/His2(NA2), Lys82(EF6) and Lys/Ser143(H21), and the marked reverse Bohr effects (φ=0.2 and 0.38, respectively) in the absence of phosphates are attributed to the close proximity of these positively charged residues in the T-state, which reduces their affinity for protons. The smaller reversed Bohr effect (φ=0.08) in cathodic Hb I of S. parasiticamay thus indicate a lower density of positive charges at the phosphate-binding site.

A significant finding is the low Cl^- sensitivity of S. parasitica Hb I (Fig. 3). Based on studies of mammalian/human Hb, two schools of thought exist as regards the molecular mechanism of the Cl^- effect: (1)oxygenation-linked Cl^- binding at specific sites(Fronticelli et al., 1994)[one between Val1(NA1) and Ser131(H14) on the α chain and another between Lys82(EF6) and Val1(NA1) on the β chain; Fantl et al., 1987; Riggs, 1988] and (2) a general neutralization of excess positive charges that destabilize the T-state in the central cavity (Perutz et al.,1994). Although the Cl^- sensitivities of abnormal human Hbs support the latter view, the Hb of the high-altitude Andean frog Telmatobius peruvianus [where loss of Cl^- sensitivity correlates with acetylation of NA1 of the α chains (as in fish) and replacement of polar α chain Ser131(H14) by nonpolar Ala] provides evidence for the implication of specific α chain sites (Weber et al.,2002). Elucidation of the primary structures of S. parasitica Hb I and Hb II promises valuable insight into the molecular basis for differentiated Cl^- sensitivities(Fig. 3).

Another striking observation is the anticooperativity(n₅₀=∼0.6) observed in the major S. parasitica isoHb at low temperature (5°C) and neutral pH (7.0)(Fig. 3B), which correlates with anticooperativity (n=∼0.64) at 30-50% O₂saturation (Fig. 4). Correspondingly low values (n₅₀=0.6) observed for CO and O₂ binding in Hbs of the deep-sea Antimora rostrata(Noble et al., 1975) and Coryphaenoides acrolepsis (R. E. Weber and F. C. Knowles,unpublished), respectively, raise the possibility that allosteric interactions may be differently expressed at atmospheric pressures compared with high hydrostatic pressures. However, no shape changes are seen in O₂-binding curves of human blood and Hb with hydrostatic pressures of 1-126 atmos (Reeves and Morin,1986). n₅₀ values below unity at low pH are diagnostic of Root effect Hbs that secrete O₂ into the swimbladders and retinae of fish (Pelster and Weber,1991). Values of <1 can result from two populations of dissimilar heme groups, which could represent different isoHbs or α andβ chains of the sets of Hbs (Noble et al., 1986). Obviously, the former possibility cannot explain the n<1 regions in the oxygenation curves of purified Hb II(Fig. 4).

The linear van't Hoff plots (Fig. 6) indicate that the enthalpy of oxygenation is temperature independent, unlike in the Antarctic fish Dissostichus mawsoni, where convex van't Hoff plots reflect increased temperature sensitivities at decreasing temperatures and marked changes in the heat capacity difference upon oxygenation (Fago et al.,1997a).

Given that S. parasitica Hb I lacks significant oxygenation-linked Cl^- binding and only shows a slight Bohr effect, itsΔ H′ in the absence of anions may be expected to approximate ΔH⁰. However, the values found (-46 kJ mol^-1 to -49 kJ mol^-1) are low compared with those for zoarcid Hbs (∼-75 kJ mol^-1) and human Hbs (-78 kJ mol^-1) (Weber,1992) at high pH, where the Bohr effect (oxygenation-linked proton dissociation) is almost zero. Curiously also, the overall heat of oxygenation of Hb I was not reduced by ATP addition (at pH 7.0, ΔH′=-44 kJ mol^-1 to -45 kJ mol^-1 in the absence and presence of ATP, respectively; not shown). Given the small variability of the intrinsic heats of oxygenation of metal-containing gas-binding proteins(Klotz and Klotz, 1955), these observations indicate the presence of other endothermic reactions that reduceΔ H′. A possible candidate is endothermic allosteric transitions, as in cathodic trout Hb I, where the low temperature sensitivity for CO binding is attributed to conformational changes conditioned to the molecules in the T-state (Wyman et al.,1977). In S. parasitica Hb II, however, the decreased temperature sensitivity is seen at high saturation(Fig. 4). In effect, this is analogous to bluefin tuna (Thunnus thynnus) Hb, where temperature insensitivity (which now reduces outward transport of heat and helps to maintain warm bodies) is attributable to the dissociation of a large number of Bohr protons late in the oxygenation process(Ikeda-Saito et al., 1983; Weber and Wells, 1989), and with tench (Tinca tinca red blood cells, where endothermic proton dissociation occurs at high saturation(Jensen, 1986). In each case,the resultant reduction temperature sensitivity will limit O₂affinity variations in the Hb in the face of variable environmental temperatures.

We thank Anny Bang for expert technical assistance and Dr Angela Fago(Aarhus) for critically reading the manuscript. We also thank the crews of the research vessels N/O Nadir and N/O Atalante and of the submersible Nautile, as well as Françoise Gaill and Daniel Desbruyères, chief scientists for the cruises during which the samples were collected. Supported by the Danish Natural Science Research Council.