The archaeogastropod mollusc Haliotis iris: tissue and blood metabolites and allosteric regulation of haemocyanin function

The New Zealand blackfoot abalone (Haliotis iris; also known by the Maori name, paua) belongs to an ancient morphologically conservative group, the Archaeogastropoda (Andrews, 1988). Blood concentrations of the oxygen-transporting protein haemocyanin (Hc) show remarkably high individual variability in Haliotis species (Pilson, 1965; Ainslie, 1980a) (H. H. Taylor and J. E. Taylor, unpublished data for H. iris). Nevertheless, oxygen-binding studies suggest that the pigment plays a quantitatively important role, transporting more than 80 % of the O₂ delivered to the tissues (Ainslie, 1980b). Ultrastructural, immunocytochemical and cDNA hybridisation studies strongly implicate rhogocytes (pore cells), located in connective tissues of the mantle, digestive gland and foot, as the site of Hc biosynthesis in Haliotis spp. and other gastropods (Sminia and Boer, 1973; Sminia and Vlugt van Dalen, 1977; Hazprunar, 1996; Taylor and Anstiss, 1999; Albrecht et al., 2001).

As in other gastropods, the Hc of Haliotis exists primarily as a didecamer of approximately 8 MDa with a hollow cylindrical quaternary structure, comprising 20 subunits of approximately 400 kDa each, arranged as two end-to-end ring-shaped decamers. However, variable proportions of decamers and multidecamers are also present (van Holde and Miller, 1995; Söhngen et al., 1997; Harris et al., 2000). Each subunit is folded into eight functional units of approximately 50 kDa (Ellerton and Lankovsky, 1983; Ellerton et al., 1983; Keller et al., 1999), each reversibly binding one O₂ molecule at a binuclear copper site (van Holde and Miller, 1995). H. tuberculata Hc exists in two isoforms (HtH1 and HtH2) (Keller et al., 1999; Lieb et al., 1999, 2000; Harris et al., 2000; Meissner et al., 2000), which correspond immunologically to the much-studied KLH1 and KLH2 isoforms of another archaeogastropod, the keyhole limpet Megathuria crenulata (Gebauer et al., 1994; Söhngen et al., 1997). Considerable progress has been made in elucidating the structure and sequence of molluscan Hcs and their genes, including those of H. tuberculata, and the evolutionary implications of these studies are currently of great interest (e.g. Gebauer et al., 1994; Cuff et al., 1998; Keller et al., 1999; Meissner et al., 2000; Decker and Terwilliger, 2000; Lieb et al., 2001; Van Holde et al., 2001).

The Hcs of marine gastropods, including Haliotis spp., commonly exhibit a pronounced reverse Bohr shift (O₂ affinity increases with falling pH) in the physiological pH range (Brix et al., 1979, 1990; Ainslie, 1980b; Petrovich et al., 1990; Wells et al., 1998). A specific adaptive advantage of the reversed Bohr shift has been proposed in relation to salinity acclimation (Buccinum undatum) (Brix and Lomholt, 1981) and to dormancy (Otala lactea) (Barnhart, 1986), but it is unclear what factors led to the appearance of this property in early gastropods. In the highly aerated sea water inhabited by these animals, the shift would tend to compromise both O₂ loading at the body surface and unloading in the tissues. Intriguingly, the reverse Bohr effect is present in shelled snails but not in other gastropods, and Redmond (1968) hypothesised that it may be related to internal conditions during their defensive withdrawal into the shell. As yet, such conditions have been poorly documented. In Haliotis spp., withdrawal consists of clamping to the substratum by contraction of the massive right adductor muscle of the shell. Both the adductor muscle and the large foot muscle are facultatively anaerobic tissues, and during functional and environmental hypoxia they accumulate d-lactate and the uncommon opine tauropine (Gäde, 1988; Baldwin et al., 1992; Wells and Baldwin, 1995). l-Lactate is an important modulator of oxygen binding to crustacean Hc (Truchot, 1980; Bridges et al., 1984; Morris and Bridges, 1986; Lallier and Truchot, 1989), but it is not known whether lactate or tauropine plays a role in oxygen delivery in Haliotis spp. Indeed, it is unclear whether these products enter the haemolymph or are retained intracellularly.

Besides pH effects, inorganic ions and CO₂ may modulate the Hc-O₂ affinity and cooperativity of O₂ binding (expressed, respectively, as the half-saturation O₂ tension, P₅₀, and Hill’s cooperativity coefficient, n₅₀), often in a temperature-dependent manner (Spoek et al., 1964; Mangum and Lykkeboe, 1979; Brix et al., 1990; Mikkelsen and Weber, 1992; Wells et al., 1998). However, compared with haemoglobins (Hbs), much less is known about the allosteric control of O₂ affinity of Hcs, and such information is unavailable for Hcs that show a reverse Bohr shift. In vertebrate Hbs, anionic organic phosphates and protons lower O₂ affinity by decreasing the O₂ association equilibrium constant of the Hb in the low-affinity, tense state (K_T) (Tyuma et al., 1971, 1973; Weber et al., 1987). In contrast, in the giant extracellular Hbs of annelids, inorganic cations and protons modulate the O₂ association equilibrium constant of the high-affinity, relaxed state (K_R) (Weber, 1981), and in the extracellular Hb of the pulmonate snail Biomphalaria glabrata protons bind preferentially to the oxygenated Hb, decreasing K_R, whereas cations bind preferentially to the deoxygenated Hb, increasing K_T (Bugge and Weber, 1999). In the Hc of the shrimp Callianassa californiensis, Mg²⁺ increases O₂ affinity by increasing both K_R and K_T, exerting a more pronounced effect on the latter (Miller and van Holde, 1974).

In this paper, we report changes in the composition of the arterial, venous and intramuscular haemolymph of H. iris during environmental hypoxia associated with prolonged emersion and modest functional hypoxia associated with repeated disturbance. To elucidate further the respiratory function of H. iris Hc and the underlying allosteric mechanisms, we measured the effects of divalent cations and of lactate in relation to the Bohr shift and cooperativity.

Male and female specimens of Haliotis iris Gmelin (approximately 200–400 g including shell) were collected around Akaroa and Lyttelton Harbours, Banks Peninsula, New Zealand, maintained in a recirculated seawater system at 15°C, on a 12 h:12 h photoperiod, and fed with commercial abalone food (AbFeed, Sea Plant Production Ltd, South Africa). Animals were acclimated to the system for at least 1 week and deprived of food for 2–3 days before experiments.

Two series of 14 animals were cannulated for sampling of either the aortic haemolymph or the interstitial haemolymph from the lacunae of the main (right) adductor muscle and transferred to 1 l containers supplied with flowing sea water from the recirculated system. After 24 h, one series (control) was left undisturbed and the other series (emersed) was drained of water without disturbing the animals. After a further 24 h, haemolymph samples (1 ml) were removed from the aortic cannula and from the pedal sinus (using a 23 gauge needle and syringe, accessed from the ventral surface of the foot in the anterior midline), followed quickly by foot and adductor muscle samples (0.5–1.0 g), from both groups. Haemolymph samples were treated immediately with 0.1 ml (1:10 v:v) of ice-cold 6 mol l^–1 perchloric acid (PCA) and centrifuged at 13 000 g for 5 min; the supernatant was frozen and stored until assayed for lactate and tauropine.

Muscle tissue samples were immediately freeze-clamped in liquid nitrogen and stored under liquid nitrogen until processed. Aortic cannulae, placed through a window in the shell, consisted of a short length of 23-gauge needle attached to several centimetres of plugged PE tubing. Adductor muscle haemolymph was aspirated from fluid which accumulated in a well (5 mm diameter and 10 mm deep) accessed via a glass tube glued into a hole in the shell drilled through the centre of the muscle insertion (N. L. C. Ragg and H. H. Taylor, unpublished method). The well was cleared 1 h before collection commenced and, in some cases, gentle suction was applied to obtain sufficient sample.

Thirteen animals in each series were sampled for both foot and adductor muscle tissues. Six of these in each series were also matched with adductor muscle haemolymph samples. Six animals in each series were sampled for both aortic and pedal blood, of which five were matched with simultaneous muscle tissue samples.

Two series of seven animals (control, disturbed) were acclimated as above (without cannulation). Control animals were removed from their containers once to quickly take a single blood sample (4–5 ml) from the cephalic arterial sinus (CAS; accessed anteriorly at the angle between the foot and the head), blotted dry and weighed. Disturbed animals were blotted and weighed (involving 2–3 min of handling and air exposure) at 30 min intervals, with a final blood sampling (4–5 ml) from the CAS at 3 h. Individual samples were centrifuged, a small sub-sample was removed for lactate determination and to measure Hc absorbance at 346 nm, and the remainder was frozen at –80°C and air-freighted on dry ice to Denmark for O₂-binding studies. For both groups, a second blood sample (0.5–1.0 ml) was taken from the CAS for P_O2 and pH determination.

Haemocyanin concentrations were estimated from absorbances at 346 nm (Uvikon 860 spectrophotometer) of fresh, centrifuged haemolymph diluted 10-fold in aerated buffer (glycine 50 mmol l^–1, EDTA 10 mmol l^–1, pH 8.8). These values were converted to mmol l^–1 HcO₂ functional units using a practical extinction coefficient (E_mM,1cm=11.42±0.17, mean ± s.e.m., N=90), uncorrected for residual scattering, derived from copper analysis (H. H. Taylor, J. W. Behrens and N. Fawzi, unpublished data).

d- and l-lactate and tauropine concentrations in the haemolymph and muscle samples were measured enzymatically. PCA-treated hemolymph samples were thawed and assayed directly. Approximately 0.5 g of the freeze-clamped muscle tissue was weighed, crushed under liquid nitrogen, then homogenised in 5 ml of 0.6 mol l^–1 PCA. The PCA extracts were centrifuged as above, and the supernatants were neutralised with 5 mol l^–1 K₂CO₃. After standing on ice for 1 h, the precipitated potassium perchlorate was removed by centrifugation, and the supernatants were stored frozen until assayed for lactate and tauropine. Metabolites were assayed spectrophotometrically following modification of the methods of Gutmann and Wahlfield (1974) and Engel and Jones (1978). Assay mixtures contained 875 μl of buffer (glycine, 333 mmol l^–1; hydrazine sulphate, 133 mmol l^–1; EDTA, 10 mmol l^–1; pH 9.0), 100 μl of 50 mmol l^–1 NAD⁺, 25 μl of PCA extract and approximately 10 i.u. of d-lactate dehydrogenase (Sigma), l-lactate dehydrogenase (Sigma) or tauropine dehydrogenase. The latter was purified from abalone adductor muscle by ion-exchange chromatography (Gäde, 1987) as modified by Baldwin et al. (1992).

In the disturbance trials, d- and l-lactate were measured in neutralized deproteinized haemolymph samples (100 μl of sample plus 200 μl of ice-cold 1 mol l^–1 PCA and 67 μl of 3 mol l^–1 KOH) using a test kit (Boehringer Mannheim no. 1 112 821). The pH of the glycyl-glycine buffer was reduced from 10 to 9, and 0 mmol l^–1 EDTA was added to reduce interference by trace metals, as above (Engel and Jones, 1978). After centrifugation, 200 μl sub-samples of the supernatants were used in the assay. All determinations were made in duplicate, and appropriate controls were run for non-specific activity. The detection limit for lactate and tauropine in blood and tissue samples was approximately 0.1 mmol l^–1.

P_O2 was measured using a Clarke-type O₂ electrode (model 1302, Strathkelvin) housed in a microcell (MC100), thermostatted at 15°C and calibrated using room air and sodium sulphite solution (for zero P_O2). pH was measured using a flat-tipped pH electrode (Activon AEP332) in a specially constructed thermostatted microcell calibrated with BDH Colorkey Buffers. Measurements were performed within 2 min of sampling.

In vivo concentrations of Ca²⁺ and Mg²⁺ were assayed on individual blood samples, diluted 20-fold, using ICP emission spectrometry (Perkin-Elmer).

Samples of 130 μl of haemolymph from individual animals were stored at –80°C and freshly thawed. Samples with varying pH were prepared by adding 1 mol l^–1 Bis-Tris buffers to a final buffer concentration of 0.1 mol l^–1. O₂ equilibria of 4–6 μl haemolymph samples were recorded using a modified gas diffusion chamber fed by cascaded Wösthoff gas-mixing pumps (Bochum, Germany) that produce stepwise increases in O₂ tension by mixing air with ultrapure (>99.998 %) N₂ (Weber, 1981; Weber et al., 1987) while absorbance was recorded continuously. Measurements of pH were carried out in parallel on 100 μl sub-samples using a BMS 2 Mk 2 microelectrode coupled to a PHM 64 Research pH meter (Radiometer, Copenhagen). All measurements were carried out at 15°C. For each O₂-binding curve, at least four equilibrium steps between 20 and 80 % saturation were recorded, and P₅₀ and n₅₀ values were interpolated from Hill plots {log[S/(1–S)] versus logP_O2, where S is the fractional O₂ saturation and P is the O₂ tension}.

Unused fractions of native haemolymph samples taken from the 14 animals used to investigate the effects of handling were pooled and dialysed against Tris buffer to remove possible cofactors (‘stripped’ Hc). Dialysis was carried out using Spectra/Por 2.1 semi-permeable tubing (molecular mass cut-off 15 000 Da; Biotech regenerated cellulose dialysis membranes) at 4°C for 24 h against three changes of 0.02 mol l^–1 Tris buffer with 0.1 mol l^–1 NaCl (pH 7.75). The individual and combined effects of Ca²⁺, Mg²⁺ and d- and l-lactate on O₂ binding were examined by adding 1 mol l^–1 CaCl₂, 1 mol l^–1 MgCl₂ and 0.5 mol l^–1 of either d- or l-lactate, singly or in combination, to obtain final concentrations of 10 mmol l^–1 Ca²⁺, 50 mmol l^–1 Mg²⁺ and 5 and 10 mmol l^–1 of the respective lactate forms.

Precise measurements emphasizing extreme (low and high) O₂ saturations (Weber, 1981) were carried out using dialysed haemolymph that had been concentrated 10-fold by centrifugation at 2600 g and 4°C for 15 h in Ultrafree-4 Millipore tubes (with 30 kDa molecular mass cut-off membranes). The data were analysed in terms of the two-state Monod–Wyman–Changeux (MWC) model, according to the equation:

where S denotes O₂ saturation, L is the equilibrium constant between the tense (T) and relaxed (R) states in the fully deoxygenated form, K_T and K_R are the association equilibrium constants for the low-affinity (T, tense) and high-affinity (R, relaxed) forms, respectively, P is the partial pressure of O₂ and q is the number of interacting binding sites (Monod et al., 1965). Curve-fitting to obtain K_T, K_R and the allosteric constant L, estimation of the standard errors and calculation of the derived parameters P₅₀, P_m (the median O₂ tension), n₅₀, n_max (the maximum cooperativity along the equilibrium curve) and ΔG (the free energy of cooperativity) were carried out as detailed previously (Weber et al., 1995).

Statistical differences were assessed using Student’s t-tests, assigning statistical significance at P<0.05.

Tissue d-lactate and tauropine concentrations in the foot and adductor muscles of abalones settled in water were low (mean values 0.4–0.8 mmol kg^–1 fresh mass) and were below the detection limit in approximately one-third of cases (Table 1) (Fig. 1). Significant haemolymph concentrations of d-lactate were present only in the adductor haemolymph samples, but tauropine levels generally similar to tissue values were recorded in haemolymph from all three sites. After 24 h of air exposure, d-lactate and tauropine concentrations were elevated in all haemolymph and tissue samples. d-Lactate was present at approximately similar concentrations in haemolymph and tissue samples (mean values approximately 1.7 mmol l^–1/mmol^–1 kg^–1, ranging up to 2.6–2.7 mmol l^–1/mmol^–1 kg^–1, respectively). Tauropine concentrations, although highly variable, were higher than the d-lactate values in both foot and adductor tissues (mean values approximately 5.5 mmol kg^–1, ranging up to 11.4 mmol kg^–1) but lower in the haemolymph samples (0.6–1.6 mmol l^–1, highest value 2.8 mmol l^–1 in the pedal sinus). l-Lactate was absent from the haemolymph and tissues of both resting and air-exposed animals (Table 1).

A simple index of the different distributions of d-lactate and tauropine was obtained from the ratio of their haemolymph concentration (either the mean of the aortic and pedal samples or the adductor haemolymph value, mmol l^–1) to the tissue concentration (mean value for foot and adductor muscle) for the 11 animals for which matched data were available. These ratios were 1.28±0.23 for d-lactate and 0.20±0.06 for tauropine. The difference is highly significant (P=0.0004, paired t-test, two-tailed).

Haemolymph Ca²⁺ and Mg²⁺ concentrations were almost identical in the control (undisturbed) and disturbed (3 h-handled) groups (Table 2). The pH was slightly, but not significantly, higher in the control group, both groups exhibiting near-neutral values (7.02 and 6.83, respectively). P_O2 values in the cephalic arterial sinus were variable and did not differ in the control and disturbed groups (20.3 and 16.1 mmHg, respectively; 1 mmHg=0.133 kPa), as also was the case with Hc concentrations (0.21 mmol l^–1 and 0.23 mmol l^–1, respectively) (see Table 2). Neither d- nor l-lactate was detected in the blood of either group.

Native haemolymph from both control and disturbed animals showed a high O₂ affinity (P₅₀=3.9 and 4.0 mmHg, respectively) and slight cooperativity (n₅₀=1.1 and 1.2, respectively) at pH 6.9 (Table 3). Increasing the pH to 7.7 significantly decreased O₂ affinity, with P₅₀ increasing to 11.6 and 12.2 mmHg for the control and disturbed groups, respectively. Cooperativity was also pH-sensitive, increasing with rising pH to 1.6 and 1.4, respectively. These data show a large reverse Bohr effect (φ=ΔlogP₅₀/ΔpH=+0.65 for the control group, and +0.68 for the disturbed group; Table 3). On the basis of the similar O₂-binding parameters (and identical haemolymph ion concentrations), the overall mean values were used for comparison with dialysed samples.

The effects of divalent cations and lactate are illustrated in Figs 2 and 3 and Table 4. Dialysis of the Hc increased O₂ affinity (decreased P₅₀ values from 11.9 to 4.5 mmHg at pH 7.7 and from 3.9 to 3.2 mmHg at pH 7.0). The greater effects at high pH resulted in a marked reduction of the reverse Bohr effect (φ falling from +0.66 to +0.21). Dialysis also reduced n₅₀ (from 1.5 to 1.0 and from 1.3 to 1.1 at pH 7.7 and 7.0, respectively).

This pronounced effect of dialysis on O₂-binding properties prompted an examination of the effects of 10 mmol l^–1 Ca²⁺ and 50 mmol l^–1 Mg²⁺ (which approximate the in vivo values) and lactate. The addition of 10 mmol l^–1 Ca²⁺ decreased the affinity at both high and low pH, and also affected the reverse Bohr factor, which increased to +0.28. P₅₀ changed more at pH 7.6, increasing from 4.5 mmHg (intrinsic affinity) to 5.3 mmHg, than at pH 7.0, and cooperativity became slightly negative (n₅₀=0.9 at both pH values). The presence of Mg²⁺ (50 mmol l^–1) markedly decreased O₂ affinity at pH 7.7 (P₅₀=8.1 mmHg), such that the pH-sensitivity of O₂ binding and cooperativity were regained (φ=+0.62 and n₅₀=1.9; see Table 4). A small increase in O₂-binding affinity was seen at pH 7.0 (P₅₀=2.9 mmHg).

The effects of both d- and l-lactate on O₂ binding of dialysed Hc were investigated at low pH (assumed to favour binding of the negatively charged ion to positively charged sites on the Hc molecule). Neither d- nor l-lactate (5 mmol l^–1) changed the affinity (at pH 7.0, P₅₀=3.0 and 3.3 mmHg, respectively), although cooperativity was slightly decreased in both cases (n₅₀=0.9 and 1.0, respectively).

To determine the possible contributions of Ca²⁺ and Mg²⁺ to Hc-O₂ affinity in vivo, the combined effects of 10 mmol l^–1 Ca²⁺ and 50 mmol l^–1 Mg²⁺ were measured and were found to lower O₂ affinity more than Mg²⁺ alone, P₅₀ increasing to 3.5 and 8.7 mmHg at pH 7.0 and 7.7, respectively. The Bohr factor was high (+0.60) and virtually unchanged compared with that with 50 mmol l^–1 Mg²⁺ (+0.62). Curiously, Ca²⁺ exerted the opposite effect compared with Mg²⁺ on cooperativity at pH 7.6, decreasing n₅₀ to 1.5 (compared with 1.9 with 50 mmol l^–1 Mg²⁺). The effects of 10 mmol l^–1d- or l-lactate in the presence of 10 mmol l^–1 Ca²⁺ plus 50 mmol l^–1 Mg²⁺ were also investigated. The additive effect of d-lactate was pH-sensitive; P₅₀ increased only at pH 7.6 (to 9.4 mmHg), resulting in a small augmentation of the Bohr factor (to +0.65). A similar pattern was found for l-lactate: P₅₀ increased to 9.5 mmHg at pH 7.6 and was largely unaffected at pH 7.0 (3.5 mmHg), such that the reverse Bohr effect remained high (+0.68) and n₅₀ decreased slightly (to 1.30) at high pH.

Precise O₂ equilibrium measurements in the absence and presence of Mg²⁺ were made at pH 7.7, where this modulator exerted a pronounced effect on the cooperativity of dialysed samples. As no cooperativity was observed in the dialysed Hc (Fig. 4) it was impossible to fit the MWC model to the data. Instead, P₅₀ and n₅₀ were determined from the linear regression to be 4.3 mmHg and 1.0, respectively. With K_T=K_R=1/P₅₀, the corresponding association constant is 0.23 mmHg^–1 (Table 5). The addition of 50 mmol l^–1 Mg²⁺ altered the affinities of both the tense and relaxed forms of the Hc molecules, the greatest effect being seen on K_T, which decreased to 0.075 mmHg^–1, while K_R increased to 0.449 mmHg^–1. The number of interacting O₂-binding sites (q) providing the best possible fit was 15.48±2.17; mean ± s.e.m.) (Table 5). The high q value together with a very high value of L reflect the narrow P_O2 range in which the molecule shifts from the tense to the relaxed state (Fig. 4). The Mg²⁺-induced variations in K_T and K_R were also obtained from analyses with q fixed at multiples of 8 (8, 16, 24 and 32, respectively; data not shown), and similar results were obtained. The P_m value was lower than the P₅₀ (7.2 mmHg compared with 8.9 mmHg), and n_max was higher than n₅₀ (3.3 compared with 2.7), reflecting asymmetry of the O₂-binding curves. The free energy of cooperativity ΔG was 4.291 kJ mol^–1 at pH 7.7.

The accumulation of d-lactate and tauropine in the muscle tissue of abalone during environmental and functional hypoxia has been reported previously and has been discussed in terms of the differing strategies available for anaerobic energy production among molluscs (Wieser, 1981; Fields, 1983; Sato et al., 1985; Wijsman, 1985; Gäde, 1988). The elevation of d-lactate levels has also been observed in the blood of pulmonate gastropods during periods of anoxia (Wieser, 1981; Wijsman, 1985), but we believe that this is the first demonstration of the presence of d-lactate in the blood of an archaeogastropod and the first published report of the presence of tauropine in molluscan blood. Similar observations have been made previously for the haemolymph of Haliotis rubra (J. P. Elias, unpublished data). The blood:tissue ratios (0.20 for tauropine, 1.28 for d-lactate) in emersed animals are uncorrected for tissue dry matter and haemolymph content. Nevertheless, they imply a rather rapid equilibration of d-lactate between the intracellular and extracellular pools and a greater retention of tauropine intracellularly. Some caution must be exercised: although the foot and adductor muscles account for a large proportion of the body mass, the release (and uptake) of these products at other sites is likely, and further studies along these lines should aim to quantify this.

The respiratory adaptations of gastropod Hc are poorly understood compared with those in arthropods. Adaptations in Hc function to different conditions may occur via changes in the pigment concentration, changes in its intrinsic structural and functional properties or changes in the type and concentration of ‘cofactor’ molecules that modulate its O₂-binding properties. Unlike vertebrates, in which organic phosphates are potent regulators of Hb-O₂ affinity, the affinity of invertebrate pigments is mainly dependent on inorganic cations (Truchot, 1975; Mangum and Lykkeboe, 1979). In contrast to crustacean Hc, in which Mg²⁺ and Ca²⁺ increase O₂ affinity (e.g. in the shore crab Carcinus maenas) (Truchot, 1975), these cations decrease O₂ affinity in Haliotis Hc. The O₂ affinity of some crustacean Hcs is, moreover, increased by l-lactate and urate (Truchot, 1980; Morris and Bridges, 1986).

The present data for H. iris Hc illustrate new modes of effector modulation of O₂ affinity. Removal of divalent cations through dialysis increased affinity and obliterated cooperativity (Figs 2, 3). In addition, the pH-sensitivity of O₂ binding (Bohr effect) was markedly reduced. The addition of 50 mmol l^–1 Mg²⁺, which approximates the in vivo value, at pH 7.7 (which falls within the physiological range in marine gastropods) (Mangum and Shick, 1972; Brix et al., 1979; Mangum and Lykkeboe, 1979) almost fully restored blood P₅₀ and even increased cooperativity beyond the level seen in native blood. The small additional effect of Ca²⁺ (at approximately in vivo concentrations) further decreased affinity and, curiously, opposed the positive effect exerted by Mg²⁺ on cooperativity, resulting in the O₂-binding properties seen in the native blood (Figs 2, 3).

Evidently, the in vivo P₅₀ is strongly dependent on the Mg²⁺ concentration. That the difference in the effects of Ca²⁺ and Mg²⁺ does not appear to be due to ion-specific binding by the Hc molecule, but simply to result from their in vivo concentration differences, is suggested by the observation (data not shown) that the two cations exerted similar effects when tested at equivalent concentrations. However, unlike Mg²⁺, Ca²⁺ did not induce cooperativity even when tested at concentrations far exceeding physiological values (data not shown), indicating a specific cation effect on n₅₀ (Fig. 3). The reverse Bohr effect favours O₂ binding at low pH (Brix et al., 1979; Mangum and Lykkeboe, 1979). Protons neutralise the negatively charged binding sites of the Hc molecule, decreasing cation binding and the effect of cations on P₅₀ and decreasing n₅₀ in the case of Mg²⁺. The cumulative effects of Mg²⁺ and Ca²⁺ in dialysed blood at low pH resulted in O₂-binding properties comparable with those in native blood.

Crustacean Hcs show highly specific but variable sensitivities to l-lactate (Truchot, 1980; Bridges et al., 1984; Morris and Bridges, 1986; Lallier and Truchot, 1989), which interacts with all four positions of the chiral carbon of the protein, thus explaining the difference in binding compared with d-lactate and other structural analogues (Johnson et al., 1984; Graham, 1985). The Hc of the whelk Busycon contrarium was reported to be insensitive to both stereoisomers of lactate (Mangum, 1983, 1992). However, the effect of lactate on the O₂ binding of Haliotis iris Hc has not previously been investigated. The almost identical, minor depressant effects of the two isoforms on affinity and n₅₀ indicate that the binding sites on Haliotis iris Hc cannot distinguish between these organic ions or that the effects reflect a more general anion sensitivity. In any case, it appears that neither form is an important modulator in vivo. However, before abandoning the concept of organic modulators of molluscan Hc, the potential role of tauropine should be examined. Unfortunately, tauropine is unavailable commercially, but methods have been published for its synthesis and purification (Sato et al., 1985, 1991).

Analyses of the precise O₂ equilibria in H. iris, represented in the extended Hill plots (Fig. 4), indicate a new allosteric control mechanism in which Mg²⁺ decreases K_T and also increases K_R, resulting in a sigmoid O₂ equilibrium curve. Hence, it increases the O₂ affinity of the oxygenated (relaxed) state but also (and proportionally more) decreases the affinity of the tense state. This differs from vertebrate Hbs, in which erythrocytic organic phosphates (such as diphosphoglycerate and ATP) and increased proton concentrations decrease O₂ affinity by decreasing K_T (Tyuma et al., 1971, 1973; Weber et al., 1987), annelid extracellular Hbs, in which divalent cations and increased pH increase O₂ affinity by increasing K_R (Weber, 1981; Fushitani et al., 1986), and pulmonate snail (Biomaphalaria glabrata) extracellular Hb, in which cations modulate K_T and protons modulate K_R (Bugge and Weber, 1999). While little is known about the control mechanisms in Hcs, the available data indicate that in the Hc of the shrimp Callianassa californiensis Mg²⁺ raises both K_T and K_R (Miller and van Holde, 1974), whereas in the Hc of the opisthobranch gastropod Aplysia limacina Ca²⁺ decreases K_T and increases K_R (Ghiretti-Magaldi et al., 1979). However, in the latter case, the Hill plots in the presence and absence of Ca²⁺ were measured at different pH values (pH 8.5 and 8.0, respectively), and the effects of cations cannot be separated from those of protons. The effect of Mg²⁺ on Haliotis iris Hc is analogous to that of Ca²⁺ on Limulus polyphemus Hc, in which this cation appears to enhance the stability of the deoxy conformation (Topham et al., 1998).

Gastropod Hc didecamers each possess 160 O₂-binding sites. Since this very large number of binding sites is not reflected by correspondingly high values of the Hill coefficient, n_H (cf. van Holde and Miller, 1995; van Holde et al., 2000), the value for the number of interacting sites (q) may correspond to the number of O₂-binding sites within the subunits that behave like allosteric entities (van Holde et al., 2000). Accordingly, it may be more appropriate to fit the standard MWC model, fixing q as the number of binding sites per subunit or multiples of these. In Octopus dofleini, the Hc dissociates into subunits upon removal of divalent cations at pH 8.0 (Miller, 1985; Connelly et al., 1989). This is also the case for Sepioteuthis lessioniana Hc (van Holde et al., 2000), in which the cooperativity expressed in the native decamer is obliterated in the subunits upon removal of divalent cations. Non-cooperativity in isolated subunits seems to be general in molluscan Hcs. In the Hc of the crustacean Callianassa californiensis, both Ca²⁺ and Mg²⁺ promote the association of subunits (Roxby et al., 1974; Miller and van Holde, 1974). Is it then the presence of divalent cations that induces distinct T and R states, or is the observed cooperativity in the presence of cations simply due to preservation of the native aggregated state? Although results with Sepioteuthis lessoniana Hc (van Holde et al., 2000) indicate that cooperativity is correlated with the presence of decameric quaternary structure rather than with Mg²⁺ concentration per se, this view is not supported by the present results. Thus, if the native aggregated state were the only prerequisite for cooperativity, then addition of either Ca²⁺ or Mg²⁺ would have exerted similar effects in establishing cooperativity. The unique correlation between the presence of Mg²⁺ and cooperativity in Haliotis iris Hc (Fig. 3) indicates either that the molecule can distinguish between these two divalent cations or that Ca²⁺ does not promote subunit aggregation.

The K_T/K_R ratio in the presence of Mg²⁺ indicates a sixfold higher affinity for the last compared with the first O₂ molecule bound to the macromolecules. This is a low factor compared with those of 313 (pH 7.4) and 68 (pH 7.4) observed in vertebrate and annelid extracellular Hbs, respectively (Tyuma et al., 1973; Weber, 1981). Similarly, the free energy (per binding site) of interaction between O₂-binding sites (ΔG=4.3 kJ mol^–1 in the presence of Mg²⁺) is low compared with that of stripped Hbs (ΔG=8.7 kJ mol^–1 for human Hb and ΔG=8.4 kJ mol^–1 for extracellular lugworm Hb) (Tyuma et al., 1973; Weber, 1981). However, given the large number of interacting O₂-binding sites (q=15.48 compared with q=4 in tetramers), the total free energy of interaction between O₂-binding sites may be large. The Hill plot (Fig. 4) indicates that the transition from the tense to the relaxed conformation occurs late in the oxygenation process, at log[S/(1–S)]=–0.2, i.e. approximately 17 % saturation, compared with approximately 8 % in human Hbs (Tyuma et al., 1973), reflecting the high degree of stabilisation of the tense conformation by internal bonds.

Recently, it has been shown that the Hc of Haliotis tuberculata exists in two isoforms (HtH1 and HtH2) (Keller et al., 1999; Lieb et al., 1999, 2000; Harris et al., 2000; Meissner et al., 2000). The two isoforms, and their component functional units (a to h), are all immunologically distinct but correspond respectively to the much-studied KLH1 and KLH2 isoforms of another archaeogastropod, the keyhole limpet Megathuria crenulata (Gebauer et al., 1994; Söhngen et al., 1997). The haemolymph concentrations of the different isoforms may vary independently (e.g. during starvation) and also differ in their tendencies to associate into didecamers and multidecamers in the presence of divalent cations. Clearly, future studies must establish the extent to which the properties of H. iris Hc reported here relate to one or both isoforms.

Repeated handling and air exposure over a period of 3 h appeared not to affect the haemolymph composition or pH (Table 1). Although the mean pH values are markedly lower than those reported previously for marine gastropods (Mangum and Shick, 1972; Brix et al., 1979; Mangum and Lykkeboe, 1979), they agree well with those of Wells et al. (1998), who found the in vivo pH of pedal sinus blood of H. iris at 20°C to be 7.39±0.2 at rest and 6.51±0.3 following 10 min of intensive exercise that caused metabolic acidosis. The fact that the reverse Bohr shift is operative in this pH range implies that O₂ unloading in the relatively acid tissues may be impaired.

Interestingly, the Hc of the gastropod Buccinum undatum acclimated to normoxic, high-salinity (35 ‰) water shows a normal Root effect (reduction in O₂ carrying capacity with decreasing pH) compared with a reverse Root effect following exposure to hypoxic conditions or lowered salinity (Brix et al., 1979; Brix, 1982). The shift was attributed to a hyporegulation of the blood ion levels, leading to a lowered Cl^– concentration and a concomitant increase in P_CO2 that lowered pH. The associated decrease in Cl^– blockage of the Hc-O₂ binding sites increases the O₂ saturation of the Hc when pH decreases and augments the effective levels of the O₂ carrier in the blood (Brix and Torensma, 1981). The combined effect of the reverse Bohr and Root shift was shown to increase the venous O₂ content significantly under hypoxic conditions, thus compensating for the reduced amount of physically dissolved O₂ in the blood (Brix, 1982).

A reverse Root shift, independent of hypoxia or low salinity, has been demonstrated in the Hc of H. iris (Wells et al., 1998). To date, no convincing hypothesis has been presented for its significance in combination with a reverse Bohr shift and associated loss of cooperativity at the lower end of the physiological pH range. The reverse Root shift may ensure an increased O₂-carrying capacity when pH drops, as inferred by Wells et al. (1998), and has been shown to be of importance in Buccinum undatum, in which animals adjusted to hypoxic conditions almost regenerate their normoxic O₂ uptake rate (Brix and Lomholt, 1981). In addition, under hypoxic conditions, the large blood volume of H. iris (approximately 52 % of wet mass) (Taylor, 1993; Ragg et al., 2000) constitutes a blood O₂ reserve that offers some protection against internal hypoxia. However, an increased affinity and loss of cooperativity at low pH do not necessarily impair O₂ unloading but only shift it to lower critical levels, where aerobiosis may supplement anaerobic metabolism. The pedal and shell adductor enzyme and metabolite profiles provide evidence that Haliotis spp. rely heavily on anaerobic metabolism during exercise or environmentally induced hypoxia (Wells et al., 1998); such results are supported by the accumulation of the pyruvate reductase end-products tauropine and d-lactate (Gäde, 1988; Baldwin et al., 1992; Ryder et al., 1994; Wells and Baldwin, 1995).

Taken together, the evidence indicates that Haliotis iris has a blood O₂-transport system in which high rates of O₂ delivery to the tissues are not a high priority. Consequently, the reverse Bohr and Root shifts do not constitute a serious impediment. Possibly, H. iris evolved under circumstances with no demanding need for O₂ delivery under hypoxia. These cold-adapted animals are found in shallow, sub-littoral, O₂-rich waters of New Zealand’s coasts, achieving their greatest abundance and largest size in the colder southern part of the South Island. The adults are mainly sessile and move only slowly when rasping. Although abalone sometimes glide rapidly to a refuge when disturbed, clamping to the substratum is their primary ‘escape’ response against predators, and they have little capacity for sustained aerobic locomotion. Intense muscle exercise is thus supported by anaerobic glycolysis. The substantial blood O₂ reserve may provide O₂ for facultatively anaerobic tissues when O₂ availability is restricted, as during clamping.

This study was supported by the Danish Natural Science Research Council (Danish Center for Respiratory Adaptation) and the Marsden Fund of New Zealand (Contract UOC 804). We gratefully acknowledge stimulating discussion and practical assistance from Dr John Baldwin (Monash University), Norman Ragg and David Just (University of Canterbury) and Hans Malte and Anny Bang (University of Aarhus).