Hypoxic Responses in a Fish From a Stable Environment: Blood Oxygen Transport in the Antarctic Fish Pagothenia Borchgrevinki

It has long been accepted that the respiratory properties of blood, notably the oxygen-carrying capacity and the shape and position of the oxygen equilibrium curve, have been responsive to evolutionary selection pressure. It is usually easy to interpret interspecific variation in respiratory properties in terms of adaptations which appear to fit an animal for its particular lifestyle, in its particular habitat. This is well illustrated by fishes, because oxygen availability in water is much more variable than in air. Active pelagic fishes, for example, living in well-oxygenated waters, tend to have oxygen equilibria favouring unloading of oxygen to the tissues, whereas sluggish fishes and those in low-oxygen conditions tend to have equilibria favouring oxygen uptake (see reviews by Grigg, 1974; Johansen & Weber, 1976; Powers, 1980, 1985; Wood, 1980; Weber & Wells, 1988).

Apart from this apparently adaptive interspecific variation, much work since the mid-1960s has shown that there is ‘sensible’, intraspecific variability as well, with both oxygen-carrying capacity and oxygen affinity being flexible components of the oxygen transport system, responsive to short-term demands. Seasonal acclimation of oxygen affinity in fish was first demonstrated in the brown bullhead, Ictalurus nebulosus, in a way that minimizes the effects of changing ambient temperature (Grigg, 1969a). Subsequently, it was found that short-term changes in blood oxygen-affinity are under the influence of controlled levels of red cell organic phosphates (RCOP) and fishes again provide good examples (Weber, 1982).

Episodes of reduced oxygen availability in aquatic environments are a common natural occurrence for many water breathers and this may lead to additional energy expenditure through raised ventilation rates (Lomholt & Johansen, 1979). When presented with hypoxic challenge, fish tend to show significant adaptive changes in haemoglobin function. Krogh & Leitch (1919) suggested that fish acclimate to hypoxia by an increased blood oxygen-carrying capacity and that, generally, animals native to hypoxia show increased haemoglobin levels. The oxygen affinity of blood as a factor limiting oxygen uptake at low ambient levels of oxygen was later implicated by Grigg (1969b). It has since been shown that blood oxygen-affinity is often increased as a short-term response to hypoxic challenge, allowing improved oxygen loading at the gills. The principal mechanism for effecting the increase in oxygen affinity involves the reduction of [RCOP] particularly as [ATP] and [GTP], These compounds exhibit specific allosteric effects on haemoglobin oxygen-affinity in fishes in addition to non-specific charge effects caused by a relative rise in intraerythrocytic pH when nonpenetrable anionic phosphate concentrations are decreased (reviewed by Johansen & Weber, 1976; Powers, 1980; Wood, 1980; Weber, 1982; Weber & Wells, 1988).

It is now accepted that the functional significance of short-term adaptive changes in [RCOP] following exposure to low oxygen levels is that they allow compensation by fishes living in environments subject to periodic deoxygenation. Although there is little doubt that fish encountering hypoxic conditions in nature do have the necessary machinery for adaptation, the difficulty is to know whether these arc special cases or, in making the response to low oxygen levels, they are simply exhibiting a response typical of their class, perhaps better developed, or perhaps not. This presents a dilemma for comparative physiologists and few, if any, observations have been carried out on similar animals from a different habitat or with a different lifestyle, for comparison. The generalization linking acclimation of P₅₀ to hypoxic exposure is based on approximately 15 species (representing 0·000006 % of the estimated number of fish species) which live either in fresh water or in specialized marine microhabitats. With the exception of trout, all these species live in habitats characterized by chronically or episodically low levels of dissolved oxygen.

Are hypoxia-induced affinity changes, then, a general or specialized feature of fishes? To answer this question, we chose to study a fish which has a comparatively low demand for oxygen and is unlikely ever to encounter natural environmental hypoxia.

Antarctic fishes in McMurdo Sound have oxygen demands among the lowest recorded for vertebrates and live in water of high oxygen content (Macdonald et al. 1987). Despite the constant water temperature of −1·8°C, the haemoglobin of antarctic fish is entirely functional and intraspecific differences in blood respiratory properties among members of the endemic Nototheniidae echo one’s expectations on the basis of lifestyle (Grigg, 1967; Tetens et al. 1984). Tetens et al. (1984) investigated the possibility of thermal acclimation of blood oxygen-affinity in the borch Pagothenia borchgrevinki but found no significant difference between the affinity of blood warmed to +4-5°C and blood from fish held at 4·5°C for 8–13 days, so no intraspecific short-term adaptability of blood oxygen-affinity by antarctic fish has been reported previously. We chose the same species for studying experimental exposure to low ambient oxygen concentration partly because of its easy availability, partly for comparison with previous work and partly because, living pelagically just under the annual ice, it is likely to encounter only high levels of ambient oxygen. P. borchgrevinki, therefore, seems to offer an opportunity for a set of ‘control’ observations.

All aspects of this study were carried out in two heated and insulated huts located on the l·5–2·0 m thick annual sea ice in McMurdo Sound, T5 km south of New Zealand’s Scott Base (77°S, 166°E) between 29 October and 14 November 1986. Electric power was supplied to the huts from a 7·5 kVA diesel generator. Transport of personnel and equipment between the ‘fish huts’ and Scott Base was by motor toboggan and sled. The larger hut had alm diameter hole in its floor, lined up over a similar hole cut through the ice with an ice auger, giving access to the ocean below. Cold running sea water was supplied by a bilge pump, for fish storage and for thermal control.

Fish measuring 150–200 mm were caught just below the ice on hand lines fitted with lures and hooks on which the barbs were flattened to reduce injuries. Only lip-hooked specimens were accepted and these were transferred from ocean to experimental tanks in the aquarium in approx. 4 s over a distance of 1·5 m. In this way the trauma of capture and live fish transport was minimal. Prior to the commencement of the experimental period, all fish were kept 3–4 days in a large holding tank through which oceanic water was circulated with a bilge pump at 2000 lh^-1. Temperature in the holding tank was −1·6 ± 0·1 °C, compared with −1·8°C in the ocean at our feet.

Thirty-four Pagothenia borchgrevinki (commonly called borchs) were distributed randomly into two groups, one in each of two 50-1 rectangular plastic tanks set into a much larger tank through which sea water was pumped. This served as a temperature-control bath, there being no flow through the experimental tanks. A 2·5 cm thick sheet of polystyrene foam floated on the surface of each of the experimental tanks to provide insulation and to simulate the reduced illumination of the snow and ice cover under which P. borchgrevinki live. The fish in both groups were postabsorptive and rested quietly on the bottom of the tank throughout the experiment, those in the control group showing a little more spontaneous activity. Throughout the exposure period, water temperatures in the experimental tanks were monitored many times daily. They ranged from −1·3°C to −1·6°C but were almost always − 1·5°C. Water in the control tank was maintained well-oxygenated by aeration. Its full saturation with oxygen (152–154 mmHg, l mmHg = 133·3 Pa, depending upon barometric pressure) was confirmed at least twice daily with an oxygen electrode. The experimental group was exposed to continuous hypoxia close to the ‘critical oxygen tension’ of 60mmHg in this species, as determined by Wells (1987). Water was lowered gradually at the start of the exposure period by the oxygen consumption of the fish until it fell to the selected value. At a -stat system with solenoid valves (developed in the Zoophysiology Department, Aarhus University) connected to a Radiometer PHM-71 analyser called for a controlled burst of aeration from an air pump fitted with large diffusers. Oxygen partial pressure within the experimental tank was monitored continuously with a Radiometer electrode connected to the analyser. A continuous record made during the period of acclimation showed that was maintained between 58 and 63 mmHg, averaging 60 mmHg. Each day, approximately two-thirds of the water in each tank was renewed with fresh sea water to prevent excessive build-up of organic wastes. In the case of the hypoxic tank, water was siphoned off and replaced with new water prebubbled with nitrogen until P_O2 was at an appropriate level, so that exposure of the fish to low oxygen level would not be interrupted. Each day pH was measured in both tanks prior to the water change. A consistent pattern emerged in which the control and the hypoxia tank pH values averaged 7·79 and 7·14, respectively, before each water change. The lower pH in the hypoxic tank derived from a buildup of CO₂ because of the intermittent aeration. It was not the result of an accumulation of organic acids, because bubbling samples of water from the hypoxic tank resulted in the pH being restored to values comparable to that in the control tank. Hence, it can be noted that the exposure to hypoxia in the experimental group was also associated with hypercapnia.

Sampling of blood commenced after 11 days and was completed after 14 days. Fish were taken from control and hypoxic tanks alternately. After spinal transection to immobilize the fish and prevent build-up of lactic acid due to struggling (Wells et al. 1984), blood samples were removed from the caudal vein as rapidly as possible, usually within 20s of removing the fish from the tank. The sample was kept chilled at all stages and subsamples were taken immediately for determination of haematocrit (14 μl), [haemoglobin] (10 μl) as described by Dacie & Lewis (1975), [ATP] (100 μl) and [lactate] (50 μl) using Sigma enzymatic test chemicals. The remainder was available for determination of the oxygen content which was done immediately using a method similar to that used previously for borch blood (Tetens et al. 1984). Samples of blood (50 μl) were transferred to disposable microtonometer tubes (Radiometer) and equilibrated to appropriate gas mixtures generated by a Wosthoff pump. Equilibration at each level of oxygenation was carried out at − 1· 5°C for 10min, the tube being vibrated vigorously on the modified clapper of a redundant firebell. Trials prior to the study had shown 10 min to be sufficient for equilibration and no haemolysis was observed. This makeshift tonometer, made necessary by the failure of the regular device a couple of days before our departure from New Zealand, proved to be highly effective. It is described elsewhere (Grigg & Wells, 1988). After each equilibration, oxygen content was determined in a Tucker chamber and the oxygen equilibrium parameter, P₅₀, and pH at equilibrium were determined at − 1·5 °C (Tetens et al. 1984).

Methaemoglobin was determined at the final stages of analysis by examining the blood in a Unicam SP-600 spectrophotometer, using extinction coefficients given in Benesch et al. (1973). In no case did methaemoglobin exceed 5%.

All the methods described are well-tested and comparable with those used in recent parallel studies on hypoxic acclimation in fishes (see next section).

At the completion of blood sampling the spleens were excised, weighed, and mass was expressed as a percentage of body mass.

Results were compared statistically using Student’s t-test. Not included in statistical analysis are a number of specimens which showed aberrant gill structures. Results from these fish are marked with separate sets of symbols on the graphs.

Studies on oxygen-binding properties of blood may suffer from potentially erroneous results when measurements are made at sub-zero temperatures. The validity of our choice of method applied to antarctic fish blood is discussed in Tetens et al. (1984), and Grigg & Wells (1988) report an evaluation of the modified tonometry system.

Sampling borchs by acute venepuncture may lead to acid-base disturbances (Wells et al. 1984), changes in blood oxygen–capacity (Cossins & Richardson, 1985), and altered haemoglobin oxygen-affinity (Soivio & Nikinmaa, 1981; Nikinmaa, 1983). Almost all the recent studies dealing with the respiratory properties of blood from fish under hypoxic challenge have employed acute sampling methods, particularly with species too small for effective cannulation. With practice, and with the additional procedure of rapid brain ablation and spinal transection, we have circumvented the problem of lactate flushes as our measurements indicate (see Table 1). Haematological indices in our control group were similar to those obtained from cannulated specimens by Tetens et al. (1984), indicating that secondary effects of stress arising from hormonal surges were minimal (see Wedemeyer & McLeay, 1981). Furthermore, the study was designed to maintain fish under wholly aerobic conditions (see Table 1), and not severe hypoxia when lactate is accumulated in the blood (see Greaney & Powers, 1978; Hughes et al. 1983). Decreased metabolic rates in hypoxic fish allow energy to be conserved (Fry, 1971; Cech et al. 1979; Jensen & Weber, 1985) and casual observations of P. borchgrevinki indicated that hypoxic fish were less active than their normoxic controls.

Measurement of nucleoside triphosphates, which do not distinguish adenine from guanosine, were assumed to consist of ATP on the basis of chromatographic separations demonstrating that the nucleoside triphosphate pool in P. borchgrevinki erythrocytes is exclusively ATP (Tetens et al. 1984).

Table 1 summarizes haematological data and respiratory properties from the blood of normoxic and hypoxic P. borchgrevinki. The normoxic values compare well with those from Tetens et al. (1984) at the same pH. The equilibrium pH (at P₅₀) was slightly higher in the hypoxic group (8·16 ± 0·05) compared with normoxic fish (8·00 ± 0·07) (P<0·05).

Exposure to a of 60 mmHg for 11–14 days resulted in a significant increase (approx. 40%) in O₂-carrying capacity. Haemoglobin concentration [Hb] rose by approximately 66% and, based on the assumption that 1g Hb combines with 1·34 ml O₂, predicted oxygen capacities of 4·94 ± 1·81 and 6·58 ± l·50 vol% for normoxic and hypoxic fish, respectively, agree well with measured capacities (see Table 1).

Haematocrit and haemoglobin concentration are correlated in Fig. 1. The linear regression equation describing the relationship in hypoxic fish was: Hct = 410[Hb] + 8·21 and the regression coefficient, r, was +0·86. The corresponding equation for normoxic controls was: Hct = 3·37[Hb] + 4-73, r = +0·80.

Thus, the mean cell haemoglobin concentration (MCHC) was lower in the hypoxic group but the difference was not significant (Table 1). There is, therefore, no convincing evidence for erythrocyte swelling, and the rise in haematocrit following hypoxic exposure apparently reflects increased oxygen-carrying capacity.

The spleens from hypoxic fish were pale in colour and expressed as a percentage of body mass were smaller (Table 1).

Hypoxic fish had higher blood oxygen-affinities than normoxic controls correlated with decreased erythrocyte ATP (Fig. 2). The relationship between P₅₀ and ATP concentration in μmol g⁻¹ haemoglobin for the blood of hypoxic borchs is described by the linear regression equation: P₅₀ = 1·84[ATP] + 13·4 and the regression coefficient, r, was +0·58. P₅₀ and [ATP] were poorly correlated for normoxic fish (Fig. 2), particularly when [ATP] >4 μmolg⁻¹. (This concentration corresponds with a phosphate: haem ratio of approximately 1:1 and, accordingly, we would not expect the allosteric role of ATP on P₅₀ to be manifested at saturation when all the ATP-binding sites on Hb are occupied.)

The left shift of the oxygen equilibrium curve following exposure to low oxygen level (see Fig. 3) is unlikely to be accounted for by the pH difference (0·16 units) when the Bohr correction of Tetens et al. (1984) is applied. Such a difference can be expected to account for only 3–4 mmHg, compared with the observed left shift of more than 10mmHg. From Fig. 3, at ⁠, the blood of hypoxic P. borchgrevinki is approximately 90 % saturated with oxygen and combines with 6·2 ml O₂100 ml⁻¹. At comparable and pH, the blood of normoxic borchs would be approximately 80% saturated and combine with 3·7 ml O2100 ml⁻¹ when interpolated from Fig. 3. The in vivo interpretation of these data requires a knowledge of blood gases and pH, which we lack at present.

Blood lactates in both groups were low and not significantly different, so we do not suspect O₂ debts from stresses.

Data from a small number of fish observed with an abnormal gill structure from either hypoxic or normoxic groups did not fall within the range of fish with normal gills. Those with gills having a white, fluffy appearance showed either aberrant haematology, or oxygen affinity that was atypical of the group. We are unable to interpret these results in terms of functional compensation. Histological sections stained with haematoxylin suggest a reduced surface area for gas exchange in fish manifesting this pathology of unknown aetiology (see Fig. 4).

The significance of our results is that the antarctic fish living in a physicochemically stable environment respond to low oxygen levels in a similar way to fish in water subject to seasonal deoxygenation. This suggests that phosphate modulation of oxygen affinity is not, strictly, an adaptive feature in the evolutionary sense, but reflects phenotypic plasticity and current utility.

Oxygen-carrying capacity increased in response to hypoxia by approximately 40% after 11–14 days. Increased carrying capacity, based on direct measurement or inferred from elevated [Hb], has also been noted in a number of fish including chronically hypoxic killifish (Greaney & Powers, 1978), trout (Swift & Lloyd, 1974; Soivio et al. 1980) and eels (Wood & Johansen, 1972). However, tench (Jensen & Weber, 1982), plaice (Wood et al. 1975) and carp (Lykkeboe & Weber, 1978), all of which are notably resistant to hypoxia, do not show significant increases in capacity when exposed in a similar way. They might occur, however, under more severe hypoxia, or within a different time frame from that shown by fishes not so resistant. Capacity responses may be very rapid, as seen in trout under acute severe hypoxia (Tetens & Lykkeboe, 1985) and after 6h of chronic exposure (Soivio et al. 1980). This suggests that capacitance responses are a rapid means of preserving oxygen delivery under hypoxic challenge, by safeguarding the oxygen diffusion gradient from blood to tissues.

The increase in oxygen-carrying capacity appeared to be the result of an increased number of red cells from the spleen. The spleen is a small organ in relation to body mass in teleosts, yet it provides an appreciable reservoir of erythrocytes which, under the control of the autonomic nervous system, may be released rapidly into the circulation (Fange & Nilsson, 1985). The spleen releases erythrocytes when fish are asphyxiated (Bonnet, 1929; Yamamoto et al. 1985), but this mechanism has not been demonstrated previously in chronically hypoxic fish. Significant differences in the mass and appearance of the spleens in P. borchgrevinki are evidence that a rapidly recruitable supply of erythrocytes is available to the hypoxic fish. The role of catecholamines in the response is poorly understood and is an obvious topic for further investigation.

On the basis of in vitro studies on the metabolism of trout erythrocytes (Tetens & Lykkeboe, 1981), Weber (1982) proposed the novel hypothesis that anoxic incubation of stored cells in the spleen might provide a source for the circulation of cells low in ATP and with a high affinity for oxygen. The hypothesis is fully consistent with our observations on borchs, where hypoxically acclimated fish had pale, contracted spleens, low erythrocyte [ATP] and high haemoglobin oxygenaffinity.

The derived parameter of MCHC in hypoxic borchs was not significantly different from that from normoxic fish (Table 1). This was not the case in chronically hypoxic killifish which had persistently high haematocrits after other haematological variables appeared to have returned to normal values (Greaney et al. 1980). The haematocrit rise in hypoxic tench, however, reflects no increase in oxygen-carrying capacity, but results from erythrocyte swelling (Jensen & Weber, 1982). Erythrocyte swelling also occurs in hypoxic carp (Lykkeboe & Weber, 1978) and in trout where, in addition, it is accompanied by a real increase in oxygencarrying capacity (Tetens & Lykkeboe, 1981).

It is possible that MCHC might decrease in P. borchgrevinki at an earlier or later stage in the process of acclimation. The significance of the observation that polar fish, including P. borchgrevinki, have very low values of MCHC in the normoxic state is unclear (Wells et al. 1980) but, for a given ratio of Hb: ATP, a lower MCHC implies raised affinity through a shift in the ATP binding constant. One may speculate that the intrinsically lower affinities of purified haemoglobins from antarctic fish (Wells & Jokumsen, 1982) compensate the low MCHC. The reduction of MCHC observed in other studies of hypoxia is an important physiological adjustment for two reasons: (i) the equilibrium between ATP and haemoglobin in an expanded cell matrix shifts the Hb towards a high oxygenaffinity state and (ii) the pH-sensitivity of the affinity (Bohr effect) is disturbed through changes in the distribution of protons across the erythrocyte membrane (Weber & Wells, 1988).

We attribute the increased oxygen affinity in hypoxic borchs to a fall in erythrocyte [NTP], perhaps assisted by a rise in intracellular pH. The drop in [NTP] is a common feature of the hypoxic response described for other fish. The Amazon catfish switches to air breathing when hypoxic, and [N I P] (principally [GTP]) decreases (Weber et al. 1979). Fish living under ice cover do not have access to air. An increase in affinity is also seen as adaptive in hypoxic aquatic breathers (Weber et al. 1975; Lykkeboe & Weber, 1978; Weber & Lykkeboe, 1978a,b; Cech et al. 1979; Kerstens et al. 1979). Eels manifest most of their hypoxic adjustment via the sensitivity of the cathodal Hb component to [GTP], and with a reduction of the phosphates, and the Bohr factor is also decreased (Weber et al. 1976). Similarly, GTP is the important modulator of affinity in carp (Weber & Lykkeboe, 1978a,b) and tench (Jensen & Weber, 1982). Plaice conform to this pattern but exhibit a different response by decreasing [NTP] when exposed to hyperoxia (Wood et al. 1975), though what is ‘normoxic’ for the flatfish may well be ‘hypoxic’ for other species. In contrast, the mechanism for increased affinity in hypoxic lampreys arises solely from the indirect influence of ATP on intracellular pH (Nikinmaa & Weber, 1984).

GTP appears to be the dominant cofactor in carp (Weber & Lykkeboe, 1978a,b), catfish (Weber & Wood, 1979) and tench (Jensen & Weber, 1982), all of which typically inhabit water which periodically becomes low in oxygen. In the case of intertidal fish (Bridges et al. 1984), Tilapia (Babiker, 1985), trout (Soivio et al. 1980), and in the antarctic fish, ATP has the dominant role in modulating the hypoxic responses of oxygen affinity (Tetens et al. 1984). Possibly, GTP exerts more effective control in fishes experiencing a wider range of aquatic oxygen concentrations.

Regulation of P₅₀ in response to hypoxia has been shown to be both rapid and graded. Within 6h of hypoxia onset, trout have halved their erythrocyte [ATP] (Soivio et al. 1980), and within lh eels showed marked decreases in both [GTP] and [ATP] (Chan, 1986). These are seen as adaptations preserving arterial oxygen content, despite a fall in The most rapid responses to hypoxia are ventilatory (Kerstens et al. 1979; Lomholt & Johansen, 1979) although immediate changes in P50 arising from catecholamine flushes have also been reported (Tetens & Lykkeboe, 1985). Increased blood oxygen-affinity in response to graded hypoxia is another feature of trout (Tetens & Lykkeboe, 1981) and Tilapia (Babiker, 1985).

Differences in the details of responses of P₅₀ and oxygen capacity may be due not to interspecific differences but to the fact that the severity or the time course for hypoxic adjustments may be insufficient to mount complete responses. This emphasizes the importance of relativity in environmental characteristics and, for fish, the attendant difficulties of comparing like with like.

The presence of only one major Hb component in P. borchgrevinki (Brittain, 1984) accords with the widely held view that Hb multiplicity permits a functional heterogeneity preserving oxygen transport under a wider variety of physicochemical conditions than could be served by a single component. No disadvantage can be seen in the single-component system of the hypoxic antarctic fish, though the tench, notable for its resistance to hypoxia, also has only one major component which, in whole blood, does not show marked acclimatory changes in Pæ but, like Hb in the high-altitude llama, may be viewed as a pre-adaptation for high affinity.

A peculiar finding in several nototheniid species in McMurdo Sound is the presence of gill abnormalities in approximately 10% of all Pagothenia and Trematomus spp. captured. The outward appearances of these gills are of pale, fluffy filaments with occluded circulation. Closer examination, however, reveals a hyperplasia of filament pavement cells (Fig. 4). The proliferation of these cells engulfs the lamellae, decreasing the surface area for gas exchange, though efferent blood vessels appear patent. P. borchgrevinki sampled from the natural population for our study included several specimens posthumously identified with the undefined pathology. Possible causes of the abnormality are parasites (which we have not detected), tumour growth (W. Davison, personal communication) or hyperplasia arising from unidentified irritants (J. T. Eastman, personal communication). Though the cause is unknown, we might have expected affected fish to fall within the hypoxic group. These data, depicted separately in Figs 1 and 2, show inconsistent responses of the blood to hypoxic acclimation and, for this reason, were excluded from statistical analysis.

The authors thank the Antarctic Division of the DSIR for logistic support, and the staff of Scott Base for outstanding support on the ice. Vilhelm Tetens is thanked for experimental advice, and Brent Beaumont for histology. This study was supported by a grant from the University Grants Committee. GCG and LAB wish to acknowledge the University of Sydney’s assistance with funds for travel to and from New Zealand.