Oxygen carriers are ancient proteins, probably evolved from enzymes that protected the organism against oxygen toxicity [1]. When multicellular organisms increased in size and complexity, their surface/volume ratios diminished, and simple diffusion of oxygen across the body wall was inadequate to reach all cells. The evolution of simple oxygen-binding proteins into multisubunit circulating proteins, in combination with the advent of circulatory systems, made the transport of oxygen from the periphery of the organism to cells possible. The evolution of the tetramer, including duplication and differentiation of the globin gene into α and β genes, formation of subunit interfaces, and quaternary-structure allosteric changes took place within a relatively short period between the branching points of hagfish and lampreys from cartilaginous fish. The separation of sharks from bony fish occurred near the α/β gene duplication [2].

In Antarctica, the chondrichthyan genera are poorly represented in the modern fauna, with only three species of sharks and eight species of skates [3], representing approx. 4% of the total fauna. The late-Cretaceous and late-Eocene fossils from Seymour and James Ross Islands are proof that many modern chondrichthyan genera inhabited the temperate waters of the Weddellian Province [4].

On the other hand, Notothenioidei are a large group of marine teleosts largely endemic to the Antarctic Ocean. Indirect indications suggest that notothenioids appeared in the early Tertiary, filling the ecological void on the shelf left by most of the other fish fauna (which became locally extinct during maximal glaciation), and began to diversify in the middle Tertiary. Reduced competition and increasing isolation favoured speciation. Notothenioids fill a varied range of ecological niches normally occupied by taxonomically diverse fish communities in temperate waters.

In this environment, the low metabolic demand and the high oxygen concentration reduce the need for Hb. The vast majority of species of the dominant suborder Notothenioidei have a single Hb, sometimes accompanied by a minor component [5]. Of the 213 species living on the shelf or upper slope of the Antarctic continent, 96 are notothenioids [6].

Why are there so few chondrichthyans in Antarctica? There is no obvious physiological reason for their scarcity. The freezing point of the blood plasma of Squalus acanthias (spiny dogfish) is −1.95 °C [7]. Since sea water freezes at −1.86 °C, this would seem to provide adequate protection against freezing, provided there is no contact with ice. Perhaps the scarcity of chondrichthyans in the modern fauna is an ecological consequence of unusual trophic or habitat conditions in the Southern Ocean [8]. In addition, the reduced diversity of teleost fishes in the Antarctic midwaters may have restricted the entry of sharks into the ecosystem.

Besides the shark species, in the Antarctic region, there are two species of Raja and six species of Bathyraja, the dominant family south of 60°S [9]. Bathyraja lacks a fossil record. The genus includes 45 extant species distributed worldwide and is supposed to have originated at least 100 million years ago [10]. Stehmann [10] proposed the following vicariance hypothesis: “During the Jurassic (213–144 million years ago), the Bathyraja lineage had an extensive latitudinal distribution in the shelf waters of most continental areas. Waters at this time were warm and homogenous throughout the world. Tectonic processes split the Bathyraja lineage into at least two stocks, one in the north-eastern Pacific and one off the South America/Antarctic component of Gondwana. At the same time waters were cooling toward the poles, and latitudinal climatic zones were developing”.

In comparing the taxonomic composition of the Arctic and Antarctic faunas, the most obvious difference is that no single group in the Arctic dominates the fauna as do the Antarctic notothenioids, where the species endemism is 88% for the benthic fauna and rises to 97% when only notothenioids are considered [11]. Species endemism for marine fish in the Arctic is 20–25% [12]. In the Arctic region, there are 26 species belonging to the class Chondrichthyes, approx. 6.3% of the total fauna.

Hbs of cartilaginous fish have been studied less extensively than those of teleosts. Only temperate elasmobranch Hbs have been investigated so far. One difficulty that presumably hindered the elucidation of structure–function relationships is the high genetic polymorphism. When the X-ray structures of two cartilaginous fish Hbs were solved [13,14], the quaternary structure and its change upon ligand binding appeared to be preserved. The proximal effect linking the quaternary-structure change to the haems was also preserved, but the distal effect, especially the role of valine E11, was altered. The Bohr-proton and organophosphate-binding sites were absent, indicating that, in such cases, the stereochemical mechanisms other than the proximal effect have evolved independently in the different species.

The present paper reports the first molecular characterization of the oxygen-transport haemoproteins of two skates living in polar habitats, namely Antarctic Bathyraja eatonii (Eaton's skate) and Arctic Raja hyperborea. The blood of the two species was found to contain two components. To assess the possible differential contribution of Hb within the evolutionary response to the two polar environments, the primary structure, the oxygenequilibrium properties and the thermodynamic features of the major component Hb 1 of both species were investigated. The primary structures of B. eatonii and R. hyperborea Hb 1 reveal significant substitutions of many residues that are generally conserved in many Hbs and in human HbA. These are likely to be related to the functional properties of these Hbs, i.e. absence of pH and organophosphate regulation, indicating that teleost and cartilaginous fish Hbs have independently evolved stereochemical mechanisms to regulate ligand binding to the haems. Moreover, similar to all cartilaginous fish, Hb 1 of B. eatonii and R. hyperborea (i) lacks helix D in the β-subunits, and (ii) has relatively low co-operativity (h≈1.6–1.8) [15]. Thermodynamic analysis indicates that the oxygenation enthalpy change (ΔH) maintains a rather constant absolute value in the pH range 6.6–8.7 in both species.

Molecular modelling was used to characterize the haem environment and to explain the lack of Bohr and organophosphate effects in comparison with the solved X-ray structures of the two cartilaginous fish [13,14].

Blood samples were drawn by cardiac puncture by means of heparinized syringes. Haemolysates were prepared as described in [16]. Purification of Hb 1 was achieved by FPLC anion-exchange chromatography, with Toyopearl Super Q-650S (B. eatonii), and Amersham Biosciences Mono P HR 5/20 columns (R. hyperborea). The Hb-containing pooled fractions were dialysed against 10 mM Hepes, pH 7.7. All steps were carried out at 0–5 °C. Hb solutions were stored in small aliquots at −80 °C until use. For oxygen binding, aliquots of a solution of CO Hb 1 were stored at −80 °C before use within a maximum of 7 days. For each experiment, one aliquot was thawed, converted into the oxy form by exposure to light and oxygen, and was used immediately. For this purpose, the Hb solution was placed in an ice bath, and the gas phase was made 100% in oxygen. With gentle agitation, the Hb was illuminated with a light source (Sylvania Model SG-50 with a DWY lamp). No oxidation was detectable spectrophotometrically in the Hb solution, indicating that final Met-Hb formation was negligible (<2%).

The overall ΔH (kcal·mol⁻¹; 1 kcal=4.184 kJ), corrected for the heat of oxygen solubilization (−3 kcal·mol⁻¹), was calculated by the integrated van't Hoff equation:

Models of α- and β-chains of B. eatonii and R. hyperborea Hb 1 were built using both the multiple alignments obtained by CLUSTALW and the single structure of D. akajei Hb, which, among the chosen sequences, shows the highest identity with the target (62% and 59% for α-chain, 56% and 58% for β-chain of B. eatonii and R. hyperborea Hb 1 respectively). Three different models for both alignments were made using the comparative modelling program MODELER [29] as implemented in InsightII (Accelrys). The zone of β-chain around β50, which corresponds to the region of removal of helix D in cartilaginous fish, was optimized further (loop modelling in MODELER). The structural validity of the models of α- and β-chains of B. eatonii and R. hyperborea Hb 1 was assessed by several criteria. A representative model from each set was selected by reference to the MODELER objective function (F, molecular probability density function violation), which describes the degree of fit of the model to the input structural data used in its construction, as well as by use of the structure verification program PROCHECK [30]. The co-ordinates of the generated models were then subjected to the Verify3D algorithm [31] using the Verify3D Structure Evaluation Server (available at http://www.doe-mbi.ucla.edu/Services/Verify_3D.html) to identify regions of improper folding. Sequence–structure compatibility was also assessed with the ProsaII program [32].

After addition of hydrogen atoms, the quaternary structures of B. eatonii and R. hyperborea Hb 1 were created using InsightII (Accelrys) and CHARMM force field for energy minimization. Energetically favourable positions for the binding of organophosphates in B. eatonii Hb 1, R. hyperborea Hb 1 and D. akajei Hb were evaluated using the docking program GROUP implemented in GRID version 21 [33]. The program GRID determines the interaction energy of a ‘probe’, representing a chemical fragment, at each point of an orthogonal grid which encloses the structure of a target macromolecule. A grid which surrounds each protein structure exceeding it by 5 Å (1 Å=0.1 nm) in each dimension and a grid spacing of 1.0 Å were selected. The standard GRID energy function and parameters were used to calculate the interaction energy. The program GROUP, by fitting the maps generated by GRID for the probes which all together represent the structure of ATP, positions the atoms of the ligand at favourable places on the protein.

The separation of the globins of B. eatonii and R. hyperborea Hb 1 was obtained by reverse-phase HPLC (results not shown). HPLC, SDS/PAGE (15% polyacrylamide) and MS indicated the presence of four different subunits in the haemolysate of both species. However, five cDNAs for α-chains and three cDNAs for β-chains were identified in B. eatonii (E. Cocca and G. di Prisco, unpublished work); in keeping with the hypothesis of genetic polymorphism, these sequences differ from one another at very few positions.

Figure 1 Amino acid sequences of the α- and β-chains of B. eatonii and R. hyperborea Hb 1

B. eatonii Hb 1 has 141 amino acid residues in the α- and β-chains; R. hyperborea has 139 amino acid residues in the α-chain and 141 in the β-chain. By comparison with the α-chain of human HbA, one deletion (αGH3) and one insertion (between αCD8 and αCD9) were found in B. eatonii Hb 1, and three deletions (αB1, αB2 and αGH3) and one insertion (between αCD8 and αCD9) were found in R. hyperborea Hb 1. The α-chains of both fish, similar to human HbA and of the temperate skate D. akajei Hb, have free valine at the N-terminus. In both β-chains, four residues are missing between helices C and E, corresponding to helix D (Figure 1).

Remarkably, in B. eatonii Hb 1, there is a serine residue at position H21 (143β), compared with histidine in mammals and lysine or arginine in other fish (except trout Hb 1, which also has serine and lacks organophosphate regulation); R. hyperborea Hb 1 has alanine at H21 (143β) (Figure 1).

The presence of histidine at NA2 (2β) in R. hyperborea is not exceptional for elasmobranchs. It also occurs in Hbs of the sharks S. acanthias [34] and Heterodontus portusjacksoni (Port Jackson shark) [35,36]. The episodic occurrence of histidine in NA2 (2β) in elasmobranchs suggests that it may be a phylogenetically primitive character.

The alignment of the polar α- and β-chains with those of HbA and of Hbs of D. akajei, the shark M. griseus and the Antarctic bony fish T. bernacchii is shown in Figures 2 and 3. The sequences of the two chains of Hbs from polar cartilaginous Hbs have comparatively higher identity with other temperate cartilaginous Hbs than with polar teleosts (Table 1). The sequence identity between Hbs of the Arctic and Antarctic skates is even higher (70–73%).

Figure 2 Sequence features of α- (A) and β- (B) chains of B. eatonii Hb 1

Figure 3 Sequence features of α- (A) and β- (B) chains of R. hyperborea Hb 1

Table 1 Sequence identities of the two chains of Hb of various fish species compared with B. eatonii and R. hyperborea

The amino acid sequences of the α- and β-chains of B. eatonii Hb 2 were also established (results not shown). The identity with the homologous chains of Hb 1 was over 95%, supporting the microheterogeneity hypothesis in cartilaginous fish Hbs.

Figure 4 Oxygen-equilibrium isotherms (Bohr effect) (A, C) and subunit co-operativity (B, D) as a function of pH of B. eatonii Hb 1 at 10 °C (A, B) and at 2 °C (C, D)

Figure 5 Oxygen-equilibrium isotherms (Bohr effect) (A, C) and subunit co-operativity (B, D) as a function of pH of R. hyperborea Hb 1 at 10 °C (A, B) and at 2 °C (C, D)

The changes in oxygen-transport protein synthesis, whether up- or down-regulation or expression of a particular combination of gene products, are often considered to be a long-term response to developmental or environmental change. On the other hand, the regulation by allosteric effectors (protons and organophosphates) is thought to be responsible for more immediate short-term perturbations.

At first glance, the p₅₀ values of polar skate Hb 1 appeared to be low when compared with some other cartilaginous Hbs, in both the absence and the presence of allosteric effectors (Table 2). However, this difference should be considered in the light of the fact that Hbs of some cartilaginous fish display the Bohr effect. Therefore, at alkaline pH, their affinity is much higher, and, at low pH, it becomes much lower [13,14]. Moreover, if the affinities are compared, taking the respective environmental temperatures and physiological pH into account, this difference becomes insignificant. As shown in Figures 4(B), 4(D), 5(B) and 5(D), co-operativity was relatively low in both species, reaching a maximum of approx. 1.8, as in all cartilaginous fish [15], indicating that a highly co-operative oxygen carrier is not needed.

Table 2 Oxygen affinity values (p50) for polar and temperate cartilaginous fish

Figure 6 Overall oxygenation enthalpy change of B. eatonii (A) and R. hyperborea (B) Hb 1

Temperature-dependence, which is governed by the associated overall enthalpy change, is an important feature of the reaction of Hbs with oxygen. Heat absorption and release can be considered to be physiologically relevant modulating factors, similar to hetero- and homo-tropic ligands. The affinity and co-operativity of all Hb systems are strictly correlated with the extent of heat absorption accompanying oxygenation. In human HbA, the apparent overall ΔH is more exothermic at alkaline pH values, where the Bohr effect is not operative and the endothermic contribution of the Bohr protons is abolished. For fish, relying upon Hbs with constant ΔH values may thus be a frequent evolutionary strategy of molecular adaptation to extreme life conditions.

A second model was built based on the single structure of D. akajei Hb; these proteins share approx. 60% sequence identity (Table 1), and, in the case of B. eatonii Hb 1, there are no insertions or deletions in the alignment. Sequence alignments are shown in Figures 2 and 3. From the three models based on the multiple alignment and the three models based on D. akajei Hb, a single representative model was chosen for each, in accordance with the MODELER objective function and PROCHECK stereochemical assessment. The stereochemistry of the selected models was acceptable in both cases. The models show no residues within disallowed regions, and only one or two residues in the generously allowed regions of the Ramachandran plot, indicating correct stereochemistry.

The overall quality of the structure was also determined by parameters that assess residue environments and atomic contacts. As shown in Figures 7(A) and 7(B), the three/one-dimensional scores of the models generated by Verify3D for α- and β-chains respectively are always positive and are similar to those obtained with the template structure of temperate D. akajei (PDB code 1CG5).

Figure 7 Verify3D plots

The reliability of the selected models was also tested by the ProsaII program, which calculates the Cβ-Cβ pair interaction energy for each residue in the sequence, producing smooth energy plots with negative values for correctly folded proteins. The ProsaII trace for candidate models of α- and β-chains of B. eatonii Hb 1 had no positive regions, indicating no misfolding, with a so-called Z-score or normalized energy values of −6.89 and −6.72 (calculated for α-chains from multiple and single alignment respectively), and −6.64 and −6.74 (calculated for β-chains from multiple and single alignment respectively), comparing favourably with those of D. akajei Hb (−5.67 and −6.04 for α- and β-chains respectively). Analogously, the Z-score values for R. hyperborea were comparable to those of Hb crystals of the temperate skate (−6.63 and −6.53 for α-chains from multiple and single alignment respectively, and −6.23 and −6.21 for β-chains from multiple and single alignment respectively).

In the proximal side of both α- and β-haems, the replacement of leucine F7 of HbA with lysine in D. akajei, B. eatonii and R. hyperborea Hb 1 (Figure 1) is worth mentioning.

In the α-haem distal cavity, glycine E3 of human HbA and D. akajei Hb is replaced by proline in B. eatonii Hb 1, and by histidine in R. hyperborea Hb 1, thus modifying the hydrogen-binding interaction pattern of distal histidine. In the β-chains, it is worth noting that two polar residues (histidine FG4 and lysine E3), involved in hydrogen-bonding with proximal and distal histidine of HbA respectively, are replaced by glycine in D. akajei Hb, B. eatonii and R. hyperborea Hb 1 (Figure 1).

By using the program GRID, we succeeded in demonstrating that ATP is indeed bound to D. akajei Hb as hypothesized [13], whereas the ATP site is absent in the polar skate Hb 1, since the replacement of Arg¹³⁰ β(H13) with leucine in both Hbs does prevent organophosphate binding. Figure 8 shows the predicted binding site for ATP in D. akajei Leu¹³⁰ β(H13) of B. eatonii and R. hyperborea Hbs, which disallows the binding within the protein cavity, is superimposed.

Figure 8 Organophosphate-binding site in D. akajei Hb

A feature which distinguishes cartilaginous from teleost Hbs is the lack of helix D in the β subunits of the former. In β-globins, it was shown previously that the presence or absence of helix D does not affect assembly into co-operative tetramers [40]. Since recombinant human Hbs engineered with β-subunits without helix D and α-subunits having helix D show a small change in oxygen affinity, the lack of helix D in the β-subunit is thought not to exert a large functional effect. B. eatonii and R. hyperborea Hb 1 do not have helix D in either α- or β-subunits, as well as the sharks S. acanthias [34], H. portusjacksoni [35,36] and M. griseus [14], and the temperate skate D. akajei [13].

The very high sequence similarity observed in B. eatonii and R. hyperborea Hbs may reflect a common origin of polar skates, but may also suggest that the primary structure in polar fish may be related to the development of cold adaptation.

Oxygen delivery to the tissues of B. eatonii and R. hyperborea occurs in the virtual absence of pH or phosphate-linked reductions in Hb 1 affinity.

These features resemble those of ancestral oxygen carriers, and also of Hbs of several amphibians, of trout Hb I and of abnormal Hiroshima Hb (histidine βHC3 replaced by aspartate), but are not unique to polar elasmobranchs. In fact, previous studies on temperate elasmobranch Hbs have shown that their functional properties do not always include well-developed co-operativity or proton and organophosphate regulation. For instance, in Torpedo marmorata Hb, the replacement of aspartate βFG4 with lysine [41] causes the loss of the salt bridge crucial for the Bohr effect in human HbA [39]; this mutation may be responsible for the lack of Bohr effect, as the Hb of Torpedo nobiliana binds oxygen with low co-operativity (h<1), and shows no appreciable pH-dependence or response to organophosphates [42]. Elasmobranchs may have diverged from the mammalian line before stabilization of the homotropic and heterotropic interactions typical of mammalian Hbs.

In contrast, some cartilaginous fish (e.g. D. akajei and M. griseus) display a modest Bohr effect and significant ATP regulation. Such regulation originated in the common ancestor of the subclass Elasmobranchii and all other jawed vertebrates; ATP was the first organophosphate regulator of Hb function [43]. The switch from ATP to 2,3-BPG regulation may have been a consequence of curtailed oxidative phosphorylation in erythrocytes of higher vertebrates such as mammals.

In summary, Hbs of polar skates appear functionally different from those of several temperate cartilaginous fish. On the other hand, the close functional similarity (also reflected in the very high sequence identity) in the two species, each living in one of the polar environments, is of interest. The Hb systems of teleosts thriving at the two poles are far from showing the sequence identities found in skates. Antarctic notothenioids, with very few exceptions, are sedentary bottom dwellers, and have a single major Hb usually displaying strong Bohr and Root effects. Among Arctic teleosts, characterized by higher biodiversity, the number of pelagic and migratory species is instead very abundant, the Hb multiplicity is higher, and the various components appear to be functionally distinct [44,45]. Although both are cold, the Arctic and Antarctic habitats differ in many aspects, e.g., in the Arctic, the range of temperature variations is wider and isolation is less pronounced. These differences appear to be minimized in the case of two benthic sedentary skates, unlikely to disperse across wide latitude and temperature gradients and living in microhabitats with similar temperatures; this similar lifestyle may have channelled the Hb functional mechanisms (Bohr effect and ATP regulation) towards coinciding evolutionary pathways. Conversely, it is conceivable that the wide environmental differences between temperate and polar waters have given rise to far more extensive differences in physiological adaptations to the respective environments than those caused by the two polar environments, which do differ from each other, but to a lesser extent.

Physiological studies provide tantalizing insights into the possible mechanisms used by polar fish in response to low temperature. The comparison of structure and function of proteins from cold-adapted and non-cold-adapted species is an additional powerful tool. In fact, it will allow us to gain insights into the extent to which strategies of cold-adaptation are similar or vary in different phylogenies, and to what extent extreme environments require specific adaptations or simply select for more generalist or phenotypically plastic lifestyles.

This study is in the framework of the Italian National Programme for Antarctic Research (PNRA), the Arctic Strategic Programme of the Italian National Research Council, the SCAR (Scientific Committee on Antarctic Research) programmes Ecology of the Antarctic Sea Ice Zone (EASIZ), Evolutionary Biology of Antarctic Organisms (EVOLANTA) and Evolution and Biodiversity in the Antarctic (EBA), and the 2003 cruise TUNU-I (Greenland).