Blood Oxygen Transport In The Free-Swimming Hagfish, Eptatretus Cirrhatus

Hagfish are primitive jawless vertebrates whose antecedents, first appearing more than 550 million years ago, are now preserved in Upper Cambrian deposits. The composition of myxiniform plasma, quite unlike that of any other vertebrates, scarcely differs from that of sea water (Robertson, 1976). In a low pressure circulation, a series of pumps propels the blood through a specialized blood system containing large venous sinuses (Johansen, 1963; Satchell, 1971, 1984). The erythrocytes contain haemoglobin, which when isolated shows an absence of cooperative oxygen binding, lack of appreciable Bohr and allosteric effects, high affinity for oxygen, and no evidence for the tetrameric association of subunits to be found in higher vertebrates (Manwell, 1963; Bannai, Sugita & Yoneyama, 1972; Li, Tomita & Riggs, 1972; Bauer, Engels & Paleus, 1975).

Manwell (1963) speculated that the blood might function as a storage system during hypoxic episodes, and it was suggested that if the blood had a transport function, then oxygen unloading must occur at remarkably low internal tensions (Johansen & Strahan, 1963). In relation to speculations on hypoxic tolerance in hagfish, it is worth noting that agnathans evolved at a time when the earth’s atmosphere contained less than one-twentieth of the present oxygen concentration (Rutten, 1970).

This paper describes, for the first time, the respiratory properties of whole hagfish blood and relates these data to in vivo blood gas tensions. Here we have recorded oxygen equilibrium curves from the blood of the polybranchiate species, Eptatretus cirrhatus, and measured arterial and venous blood gas tensions in both resting and swimming hagfish.

Hagfish were collected in baited crayfish pots off Motunau Beach, North Canterbury (43° 05′ S, 173° 06′ E) and were held in a mesh trap prior to transfer to Christchurch. Here they were housed in aquaria containing running sea water at 16°C, the temperature at which all experiments were performed. Animals were held for at least 1 week before use and were not fed during this period. The hagfish ranged in weight from 628 to 914 g.

Eight hagfish were anaesthetized in a solution of 0·04% Benzocaine in sea water. Under anaesthesia, a cannula was placed into the 6th or 7th afferent branchial artery on one side for the collection of venous blood. Arterial blood was collected through a second cannula inserted into the dorsal aorta via a segmental artery close to the vent. Polyethylene cannulae (Portex) were used, of internal diameter greater than 1 mm, and they were filled with a heparinized saline based on the analyses of Myxine serum (Morris, 1965). Unless otherwise stated, hagfish were allowed to recover for at least 16 h before experimental manipulations and blood sampling.

Approximately 0·5 ml of blood was collected into a heparinized tube and placed on ice. Haematocrit (Het) was estimated by centrifugation at 3250g for 6 min and haemoglobin (Hb) concentration determined spectrophotometrically from Drabkin’s cyanmetHb derivative (ICSH, 1978) with the added precaution of centrifuging the extract to remove particulate matter. A relative molecular mass of 16000/monomer was assumed. Erythrocytes were counted in a Neubauer chamber filled with a suspension of cells at a 1:50 dilution. Mean corpuscular haemoglobin concentration (MCHC) was calculated as Hb concentration × 100/Hct.

Plasma lactate, erythrocyte ATP (non-specific nucleoside triphosphate), and 2,3-diphosphoglycerate concentrations were determined spectrophotometrically using the enzymatic test procedures of the Sigma Chemical Co.

Hagfish were forced to swim in a flume at 20 cms^-1 (approx. 0·4 body lengths s^-1). The in inspired water did not fall below 137 mmHg. Blood samples were withdrawn after 10–15 min swimming.

Blood gas tensions and pH were measured in heparinized blood using Instrumentation Laboratory and Radiometer electrodes thermostatted at 16 °C and connected to IL and Radiometer gas monitors. The pH electrode was calibrated using precision buffer solutions and the gas electrodes with certified mixtures. Stable readings were obtained after 4 min. Blood oxygen contents were measured using Tucker’s (1967) method.

Blood buffering capacity was estimated by tonometering 3 ml of blood at 16°C at known > with > 140mmHg or in an IL237 tonometer. Samples of 50 μl were taken for measurement of pH and total CO₂ using the method of Cameron (1971). The solubility of CO₂ in hagfish blood was taken to be the same as that in sea water at the same temperature, and dissolved CO₂ was subtracted from the measured total CO₂ to give the non-bicarbonate buffering capacity of whole blood. Regression lines were fitted to the data by the method of least squares.

Whole blood oxygen equilibrium curves (OEC) were generated from thin blood films using the dual wavelength optical system of an Aminco Hemoscan Analyzer and modified for gas supply with Wôsthoff gas mixing pumps (Wells & Weber, 1985). Dynamic error was estimated from static conditions whereby gases were admitted to the apparatus in a stepwise manner enabling the blood film to come into equilibrium with the gas. Dynamic error was found to be negligible at the slow rate and extreme pH ranges at which the OEC were registered. The CO₂-Bohr effect (ø = Δ logP₅₀/Δ pH) was determined by varying the in the equilibrium gases. The equilibrium pH was measured following tonometry for 10–12 min in a Radiometer BMS2 system with a gas blend closely matching the P50 value determined for each oxygen-binding curve. The possibility of a Root effect (reduction in O₂-binding capacity at low pH) was investigated by monitoring, in the Hemoscan, the saturation of a blood film equilibrated consecutively with pure O₂, 5 % CO₂ in air, and finally O₂. A haemolysate was prepared by osmotic shock, stripped of organic phosphates, and the OEC measured as described earlier (Wells, 1982). OEC data were analysed according to Hill’s (1910) equation. Detailed Hill plots were constructed by computer using digitized data pairs and analysing slopes at 20, 50 and 80% saturation.

All blood films were analysed after equilibrium runs to check for oxidation to methaemoglobin using a Pye Unicam SP 1750 recording spectrophotometer and the extinction coefficients given by Benesch, Benesch & Yung (1973). All data are expressed as mean ± S.E.M. and significances were calculated from Student’s t statistic, where appropriate.

At rest, the of dorsal aortic (DA) blood was 90·5 ± 6·5 mmHg (mean ± S.E.M., N=8) and ventral aortic (VA) was 16·8 ±3·1 mmHg (N = 7). For dorsal and ventral aortic blood samples taken within a few minutes of each other, the arteriovenous difference averaged 0·25 pH units (DA 7·92 ± 0·01 mmHg, VA 7·67 ± 0·07) and was 0·7mmHg lower in arterial blood (P<0·05, N =6). In three cases where it was possible to sample VA and DA blood successively from resting, swimming and recovered hagfish, it was found that was maintained during swimming and even rose during recovery (Table 1). The of venous blood returning to the gills fell dramatically (Table 1) and was not restored to pre-swimming levels even after 20 min recovery.

The hagfish swam well at the low velocities tested and were not exhausted by 15 min of exercise. Measured lactate concentrations in VA and DA blood taken after 20 min recovery averaged only 0·43 ±041 mmol I^-1, compared with the preswimming value of 0·35 ± 0·06 mmol l^-1(N = 6). By contrast, considerably higher blood lactate concentrations of 2·7 and 74 mmol I^-1 were measured in two other specimens 3 h after anaesthesia, where ventilation had been suppressed. These two high blood lactate concentrations were associated with a marked acidaemia with at 7·46 and 7·49, respectively.

Oxygen contents were lower in ventral aortic than in dorsal aortic blood (1·09 ± 0·08 mldl^-1 at of 18·9 ± 3·2 mmHg and 2·22 ± 0·15 mldl^-1 at of 82·3 ± 1·46mmHg, N=5). The arterial blood was almost fully saturated with oxygen, being close to the measured blood oxygen capacity of 2·29 ml dl^-1 at a mean haematocrit of 8^·5 %. If we assume that plasma has an oxygen capacitance identical to that of sea water, it can be calculated that 15·3 % of the total oxygen is carried in solution at the high found in arterial blood.

Haematological measurements for dorsal aortic blood are given in Table 2. Erythrocyte ATP concentrations varied from 4·53 to 7·59μumolg^-1 haemoglobin and no 2,3-diphosphoglycerate was detected. Erythrocytes were few in number (210–280× 10³μl^-1) and large in size (446–555 fl). It was not possible to measure all variables in all fish at the same time.

The in vitro buffer lines showed a low buffering capacity and a significant Haldane effect (Fig. 1), with values for deoxygenated and oxygenated bloods averaging 4·6 and 3·5slykes, respectively.

The OEC of hagfish blood under standard conditions at 16°C and pH = 7·80 showed a moderately high oxygen affinity, P₅₀ = 12·3 ± 0·55 mmHg, with mild cooperativity, Hill’s n = 1·38 ± 0·08 (N = 5). The oxygen affinity was increased in the purified haemoglobin (Fig. 2). The pH sensitivity of the equilibrium is shown in Fig. 3, where the Bohr factor, ø, — ΔlogP₅₀/ΔpH = –0·43 and the pH sensitivity is greater at P₅o than at P₂₀ or P₈₀-Hill’s cooperativity coefficient, n, was independent of saturation and thus the average n values were recorded as a function of pH (Fig. 3). There was no evidence for a Root effect.

Arterial in both resting and swimming hagfish was maintained sufficiently high to ensure virtually full saturation, with oxygen tensions close to 100 mmHg and a relatively high blood oxygen affinity. This finding contrasts with the earlier estimate of less than 50 % arterial saturation at mmHg in an eptatrid hagfish (Manwell, 1963) but corresponds with high arterial saturations in lamprey cyclo-stomes (Johansen, Lenfant & Hansen, 1973). Similarly high arterial was observed in the swimming shark Negaprion (Bushnell et al. 1982) which has blood O₂-binding properties comparable with those of E. cirrhatus. Oxygen tensions in mixed venous blood from the ventral aorta was, by contrast, low in resting fish and even lower in swimming fish (3·5 ± 1·6 mmHg). These measurements, taken together with the low estimates of plasma lactate, confirm that the blood has an important role in transporting oxygen to the tissues both at rest and during swimming. At rest, hagfish have very low metabolic rates (Smith & Hessler, 1974), though the limits to aerobic scope for activity have not been determined.

The haematocrit, haemoglobin concentration and oxygen-carrying capacity of the blood was low compared with measurements from lampreys (Beamish & Potter, 1972) and other fish species when expressed on a unit volume basis. Dissolved oxygen is a significant component of the total blood oxygen content and the high resting arterial of Eptatretus blood will increase this fraction.

The blood of E. cirrhatus has a relatively high affinity for oxygen (P₅₀ = 12·3 ± 0·6mmHg at pH 7·8 and 16°C), exhibits weakly cooperative oxygen binding (Hill’s n = 1·38 ± 0·08 at pH 7·8 and 16°C) and has a modest CO₂-Bohr effect (ø = –0·43). At approaching 100 mmHg, i.e. that obtained in the dorsal aorta, the haemoglobin will obviously be saturated with oxygen. This conclusion differs from that of Manwell (1958) who found low saturation and in the dorsal aorta of E. stouti. It should be pointed out, however, that he collected blood from a ‘sliming’ fish nailed to a board, and this cannot be related to any natural physiological situation.

In our study, the resting of 17·2 ±7·0 mmHg at pH 7·77 corresponds to a saturation of approximately 58% from the appropriate oxygen equilibrium curve. The unloading tension in the resting hagfish is therefore close to the P₅₀. Venous falls further during swimming and as a consequence of the modest degree of cooperative oxygen binding, a further 40 % bound oxygen may be released at the prevailing pH of 7·7.

The oxygen affinity, cooperativity and Bohr effect in whole blood from E. cirrhatus are strikingly similar to equilibrium data from elasmobranchs such asthe dogfish, Squalus acanthias (Wells & Weber, 1983) and the carpet shark, Cephaloscyllium isabella (Tetens & Wells, 1984). The P₅₀ and Bohr factors were also similar to values obtained for whole blood in the lamprey, Lampetra fluviatilis (Nikinmaa & Weber, 1984). There are, surprisingly, no published data of cooperativity coefficients from whole blood in lampreys or hagfish, although buffered erythrocyte suspensions from the lampreys, Geotria australis and L. fluviatilis yielded n values which rose from close to 1·0 at low saturation to 2·0 and 1·5 at 80% saturation (Macey & Potter, 1982; Bird, Lutz & Potter, 1976). In this respect, hagfish blood differs because cooperativity is less marked and the linear Hill plots indicated that n values were independent of saturation. As in the lamprey erythrocytes (Bird et al. 1976), cooperativity in hagfish blood was independent of pH over the range tested.

Hyperbolic oxygen-binding curves of relatively high affinity are an expected feature of fish with either a low demand for oxygen or with poor access to oxygen in the environment, and thus oxygen transfer may be effective over a relatively wide range of values. Highly active fish with good access to oxygen tend to have sigmoidal oxygen-binding curves and lower affinities (Riggs, 1970; Powers, 1980). A particularly high affinity for oxygen and n value of 1·1 in the tuna (Cech, Laurs & Graham, 1984) appears to be an extraordinary exception.

The role of phosphorylated metabolic intermediates in the erythrocytes of hagfish is not known. The principal phosphate isolated from Eptatretus stouti was ATP (Bartlett, 1982), as we have found for E. cirrhatus. A clue to the possible significance of erythrocyte organic phosphates in modulating haemoglobin function in the Cyclostomata is known only from the study of Lampetra, in which an indirect role of ATP in intra-erythrocytic pH was demonstrated, rather than an allosteric role (Nikinmaa & Weber, 1984). In our hagfish, the stripped haemolysate had a markedly higher affinity than that of whole blood at the same pH (Fig. 2), which might also reflect a relatively lower intra-erythrocytic pH. Hagfish haemoglobins appear to be monomeric (Li et al. 1972; Paleus & Liljeqvist, 1972) and thus the presence of a Bohr effect and weak cooperative oxygen binding is unusual. Our demonstration of a Bohr effect and weak cooperativity in whole hagfish blood is corroborated by the haemolysate studies of Bauer et al. (1975) who further found that CO₂ appeared to increase the cooperativity of stripped lysate from Myxine.

The blood of Eptatretus is poorly buffered in comparison with elasmobranch and teleost blood (Albers, 1970; Eddy, 1976; Heisler, 1980; Albers, Gotz & Welbers, 1981), which presumably reflects primarily the low concentration of haemoglobin. However, at rest DA blood is relatively alkaline when compared to elasmobranch and teleost blood at the same temperature, although lying within the range of values compiled by Heisler (1984). Using the mean pH values of VA and DA blood given above, from the fitted regression lines for oxygenated and deoxygenated blood (Fig. 1), it can be calculated that the Haldane effect is responsible for the loss of from 1·5 to 2·4 mmol l^-1 of bicarbonate as CO₂ at the gills, or from 14 to 22% of total bicarbonate. This represents a substantial Haldane effect when the low haemoglobin concentration is considered.

The hagfish Eptatretus cirrhatus occurs in southern cooler waters to depths of over 1000 m (Ayling & Cox, 1982) but may also occur in shallow water. The habits of hagfish are poorly known, although it is claimed that myxinoids inhabit muddy burrows, whereas eptatrids are frequently found on rocky substrata (Hardisty, 1979). Our underwater field observations indicate that the animals prefer to inhabit rocky crevices and that they are not seen in sandy or muddy sites where other burrowing forms are observed. Myxine survives well with the nostril occluded (Johansen & Strahan, 1963) and uses the skin as the major respiratory exchange surface (Steffensen et al. 1984). Being a considerably larger animal, the lower surface area to volume ratio of E. cirrhatus will not favour transcutaneous oxygen uptake. When feeding within the body of dead or dying fish (Hardisty, 1979), gill oxygen uptake might be impaired and at such times the blood might function as an oxygen store. The comparatively high affinity of its blood for oxygen and weakly cooperative oxygen binding are features which may facilitate oxygen transfer in an animal with a very low metabolic rate and low internal oxygen pressures. These properties, together with a weak Bohr effect, apparently enable the blood to transport oxygen and CO₂ both at rest and when swimming. Our demonstration of appreciable venous desaturation during aerobic activity shows that E. cirrhatus exploits the full range of its equilibrium curve in a blood circulation with large venous sinuses and allows for an efficient uptake of oxygen at the gill surfaces.

The authors thank D. Tattle for collection of specimens and for underwater observations. We also thank an unknown referee for valuable comments.