ABSTRACT
The abundance, distribution and oxidative capacities of mitochondria have been investigated in the red pectoral fin adductor muscles of fish (Order Perciformes) that use a predominantly labriform style of swimming. Mediterranean Sea species from the families Labridae, Serranidae, Sparidae and Antarctic Nototheniidae and non-Antarctic Nototheniidae and Channichthyidae were studied. Sub-Antarctic species from the Beagle Channel, Tierra del Fuego, included the pelagic haemoglobin-less icefish (Champsocephalus esox) and the róbalo (Eleginops maclovinus), which occurs as far north as 35°S. In Champsocephalus esox, the mitochondrial volume density of red muscle was 0.51 and mitochondrial cristae surface density (43.9 μm2μm−3) was higher than reported for Antarctic icefishes. In the red-blooded, active pelagic or semi-pelagic species, mitochondrial volume density was within the range 0.27–0.33 regardless of habitat temperature. Amongst less active demersal species, mitochondrial volume density ranged from 0.29–0.33 in polar species to 0.08–0.13 in Mediterranean species. In Antarctic species and Champsocephalus esox, myofibrils occurred in ribbons or clusters one fibril thick entirely surrounded by mitochondria. The volume density of intracellular lipid droplets was not correlated with activity patterns or habitat temperature. In a comparison of Eleginops maclovinus caught in summer (approximately 10°C) and winter (approximately 4°C), mitochondrial volume density did not differ, whereas the surface density of mitochondrial clusters was higher in summer fish. The temperature-dependence of the state 3 respiration rate of isolated mitochondria with pyruvate as substrate was described by a single quadratic relationship for all species, indicating no significant up-regulation of the maximum rate of oxygen uptake per milligram mitochondrial protein in Antarctic species. Our results support the conclusion that increasing the volume and surface density of mitochondrial clusters is the primary mechanism for enhancing the aerobic capacity of muscle in cold-water fish.
INTRODUCTION
Antarctic fish are archetypal stenotherms restricted to a narrow range of temperatures (−1.86°C to +2°C) that would be lethal for temperate and tropical species. There is a long history of studies on the mechanisms underlying cold survival and on the ecological consequences of low temperature for patterns of activity, growth and reproduction in polar fish (for recent reviews, see Johnston, 1989; Hubold, 1991; Clarke and Johnston, 1996). Ege and Krogh (1914) predicted that cold-water ectotherms would have higher metabolic rates then temperate species after making a suitable correction for differences in body temperature. Early experimental work appeared to confirm this prediction for Arctic (Scholander et al. 1953) and Antarctic (Wohlschlag, 1960, 1962) fish, and the concept of ‘metabolic cold adaptation’ (MCA) became established in the literature. Holeton (1973, 1974) subsequently reported much lower routine metabolic rates for Arctic fish, pointing out that handling stress may have contributed to the elevated levels of metabolism reported by early workers. However, the existence of metabolic cold-adaptation in polar fish continues to be the subject of investigation and controversy (Johnston et al. 1991; Steffensen et al. 1994) and has been offered support by some studies (Wells, 1987; Torres and Somero 1988). Unfortunately, the MCA hypothesis cannot be tested experimentally since the metabolism of warm-water species cannot be determined at polar temperatures. In an attempt to overcome this problem, Crockett and Sidell (1990) measured mitochondrial enzyme activities in homogenates in order to compare temperate and tropical species directly. At an assay temperature of 1°C, they found that citrate synthase and cytochrome oxidase activities in red muscle were 1.5-to fivefold higher in Antarctic than in tropical species, which is consistent with some expansion of aerobic metabolism. More recently, only modest adjustments in the maximal rates of mitochondrial respiration per milligram protein were reported in fish living at different temperatures. For example, the rates of pyruvate and palmitoyl carnitine oxidation at −1°C for red muscle mitochondria from the Antarctic fish Notothenia coriiceps were similar to the estimated rates at this temperature for muscle mitochondria from a thermophilic species living at 42°C (Johnston et al. 1994).
A common response to cold-acclimation in eurythermal fish is an almost twofold increase in the abundance of muscle mitochondria (Johnston and Maitland, 1980; Egginton and Sidell, 1989; Sänger, 1993). Various adaptive explanations include the hypotheses that increases in mitochondrial volume density partially compensate for the reduced catalytic capacity at low temperatures (Johnston, 1982a; Egginton and Sidell, 1989) or alternatively offset the reduced diffusion coefficients of cytosolic metabolites (Tyler and Sidell, 1984; Sidell and Hazel, 1987). Red muscle fibres in some polar fish habitually living at cold temperatures also have very abundant mitochondria, and similar adaptive explanations have been offered (Johnston, 1987; Londraville and Sidell, 1990). Indeed, mitochondrial volume density in the red myotomal muscle of the Antarctic silverfish Pleuragramma antarcticum (0.56) is amongst the highest recorded for vertebrates (Johnston et al. 1988).
Studies on aerobic metabolism and body temperature have often compared unrelated species with diverse lifestyles and modes of locomotion. Systematic variations in aerobic metabolism at the whole animal, organelle or enzyme level are probably to be expected both between distantly related genera and among species with different locomotory habits (Garland and Carter, 1994). The Order Perciformes show an extensive radiation, inhabiting a wide latitudinal range and both marine and freshwater environments. Species from the sub-order Notothenioidei are dominant components of the icthyofauna of the Southern Ocean, the Patagonian shelf, the Beagle Channel and the coastal zones of Argentina and Chile (Eastman, 1993; Pequeño, 1989). Notothenioidei species in the Beagle Channel, Tierra del Fuego, include Champsocephalus esox, the only haemoglobin-less icefish found outside the Antarctic, and Eleginops maclovinus, the single representative of the sub-family Eleginopinae which is absent from the Antarctic and distributed as far north as Uruguay (35°S) (Eastman, 1993). Eleginops is thought to have separated early during the radiation of nototheniid fishes and to be a phylogenetically primitive form related to Dissostichus (Balushkin, 1984). Its present-day occurrence may reflect a historical distribution on part of the Gondwana shelf that did not become associated with the margins of the Antarctic plate (Eastman, 1993). Thus, the properties of this species are of particular interest as it would have been relatively little influenced by large-scale tectonic movements or the subsequent cooling of the Southern Ocean during the Tertiary period.
The aim of the present study was to use appropriate species comparisons to test the hypothesis that increased mitochondrial abundance provides the main mechanism for adjusting muscle aerobic capacity in cold environments. We therefore determined the abundance and oxidative capacity of red muscle mitochondria in perciform species from the families Labridae, Serranidae, Sparidae, Channichthyidae and Nototheniidae that use a similar labriform style of swimming at sustainable speeds, encompassing demersal, semi-pelagic and pelagic species.
MATERIALS AND METHODS
Fish
All the fish species studied belonged to the Order Perciformes. Further taxonomic information and details on the habit, number, standard length and body mass of the fish used are given in Figs 1 and 2. The Antarctic fish Lepidonotothen nudifrons (Lönnberg) and Trematomus newnesi (Boulenger) were caught around the South Orkney Islands (60°43′S, 45°36′W) during austral summer 1994–95 (sea temperature −1.0 to 0°C). Fish were transported to the UK by the British Antarctic Survey and maintained in a purpose-built cold-aquarium at the Gatty Marine Laboratory (St Andrews, Scotland) for several months at approximately 0°C. Antarctic fish were fed on a diet of krill and squid. The sub-Antarctic fish studied were caught using trammel nets and traps in the Beagle channel in the vicinity of Ushuaia, Argentina, during 1994, 1995 and 1996. The temperature records for the fishing sites used in the study are shown in Fig. 3. Róbalo (Eleginops maclovinus Valenciennes) was caught in February (summer) and August (winter). Dorado (Paranotothenia magellanica Forster), lorcho (Patagonotothen tessellata Richardson) and the haemoglobin-less pickerel icefish Champsocephalus esox Günther (family Channichthydiae) were sampled in February. All fish were used within 48 h of capture. Coris julius L., Lithognathus mormyrus L. and Serranus cabrilla L. were caught in the Bay of Naples, Mediterranean Sea, during the autumn of 1996 (temperature 18–20°C). All the fish were sampled within a few days of capture. A further five Serranus cabrilla were obtained for experiments on mitochondrial oxygen consumption in the spring of 1997. They were transported to the Gatty Marine Laboratory and were maintained at 20°C in a recirculating aquarium for 2 weeks and fed a diet of chopped squid and shrimps.
Tissue sampling
The pectoral fin adductor muscle of the species studied is composed of a mixture of muscle fibre types. All species contained a large region of homogeneous red muscle fibres adjacent to the pelvic girdle (J. Calvo, I. A. Johnston, D. Fernandez and C. E. Franklin, in preparation). For ultrastructural analyses, small bundles of red muscle fibres were isolated from the red muscle zone and pinned via their insertions to cork or Sylgard strips and immersion-fixed in 2.5% glutaraldehyde, 0.15 mmol l−1 phosphate buffer, pH 7.4, at 4°C. Samples were subsequently cut into 3–5 small pieces, post-fixed in phosphate-buffered 1% (m/v) osmium tetroxide, rinsed, dehydrated and embedded in Araldite CY212 resin (EM-scope, Ashford, England).
Quantitative ultrastructural analysis
Isolation of mitochondria
Mitochondria were isolated at 4°C from the pooled red portion of the pectoral fin adductor muscle dissected from both sides of the body. Mitochondria were isolated as described in Johnston et al. (1994) in 8 volumes of a medium containing (in mmol l−1): 140, KCl; 10, EDTA; 5, MgCl2; 20, Hepes; 0.5% (m/v) bovine serum albumin (BSA), pH 7.4 at 4°C. Tissue was homogenized with a motorized Potter–Elvejhem tissue grinder (two passes with a loose pestle and two passes with a tight pestle, both at 900 revs min−1). Mitochondria were prepared from the homogenate by differential centrifugation (1400 g for 5 min and 9000 g for 7 min). The pellet was resuspended in the respiration medium, which contained (in mmol l−1): 140, KCl; 5, Na2HPO4; 20, Hepes, 0.5% (m/v) BSA, pH 7.4 at 4°C. The temperature-dependence of the pH of the assay medium was −0.015 pH units°C−1. A sample of the mitochondria was resuspended in the assay medium minus BSA, washed by centrifugation (9000 g for 10 min) and resuspension, and the protein concentration determined in duplicate using a modified Lowry method, using 10% (m/v) deoxycholate to solublise membranes. Biochemicals were obtained from Sigma Chemicals (St Louis, USA).
Oxygen consumption measurements
The rate of oxygen consumption of mitochondria was measured at a range of temperatures using a water-jacketed respiration cell fitted with a Clarke-type electrode (Rank Brothers Ltd, Cambridge, UK). The system was calibrated with the respiration medium saturated with air over the same temperature range using a couloximeter (Johnston et al. 1994). Approximately 0.5 mg of mitochondrial protein was added to 1 or 1.5 ml of assay medium. The rate of oxygen uptake was measured following the addition of 0.1 mmol l−1 malate to spark the Krebs cycle and either 2.5 mmol l−1 pyruvate or 0.025 mmol l−1 palmitoyl-DL-carnitine. The maximal state 3 rate was obtained by adding 0.5 mmol l−1 ADP. After a short initial period, state 3 respiration rates were approximately linear until the ADP was exhausted. The respiratory control ratio (RCR) was calculated from the ratio of the state 3 rate (+ADP) to the state 4 rate determined after all the ADP had been phosphorylated (Estabrook, 1967).
Statistics
All results are expressed as means ± S.E.M. Ultrastructural data were compared using three-way analyses of variance (ANOVAs) with activity pattern and geographical zone as between-subject factors and species as a nested factor with geographical zones. The species were classified a priori as either highly active (T. newnesi, Champsocephalus esox, E. maclovinus, Paranotothenia magellanica and Coris julius) or moderately active/sluggish (Lepidonotothen nudifrons, Patagonotothen tessellata, Lithognathus mormyrus and S. cabrilla). Further post-hoc Fisher’s protected least significant difference tests were carried out. A second-order polynomial was used to regress the rate of oxygen uptake of isolated mitochondria versus assay temperature.
RESULTS
Muscle ultrastructure
The quantitative ultrastructural measurements of red pectoral fin adductor muscle fibres are shown in Table 1. Most species had red muscle fibres with mean diameters of 17–23 μm regardless of body size or habit (Table 1; Figs 1A–E, 2A–D). Larger mean fibre diameters were found in Champsocephalus esox (56 μm), Patagonotothen tessellata (41 μm) and Lepidonotothen nudifrons (37 μm). The total volume density of mitochondria, Vv(mt,f), was 0.51 for Champsocephalus esox and between 0.27 and 0.33 for all the red-blooded sub-Antarctic and Antarctic species studied (Table 1). No consistent differences were noted among the mitochondrial volume densities of active mid-water species (Trematomus newsnesi, 0.31; Eleginops maclovinus, 0.31–0.33; Paranotothenia magellanica, 0.27) and a relatively sedentary demersal species Patagonotothen tessellata (0.33) (Table 1). Of the Mediterranean species studied, only red muscle fibres in Coris julius had comparable mitochondrial volume densities to the polar species, with significantly lower values found in Lithognathus mormyrus (0.13) and Serranus cabrilla (0.08) (Table 1). The three-way ANOVA revealed highly significant main effects of activity pattern (F(1,178)=50.2; P<0.0001) and species (geographical zone) (F(5,178)=26.8; P<0.0001) on mitochondrial volume density. The percentage of the total mitochondria found in the sub-sarcolemmal area varied from 37.9% in Champsocephalus esox to 76.4% in Lithognathus mormyrus (Table 1) and showed no consistent trend with geographical region (zone) or activity patterns. There were significant differences in the surface density of mitochondrial clusters between activity patterns (F(1,178)=35.5; P<0.0001) and between species (geographical zone) (F(5,178)=6.17; P<0.0001), but not amongst geographical zones. The volume density of intermyofibrillar mitochondria in Champsocephalus esox (0.31) was 2–3 times higher than that of the red-blooded Antarctic and sub-Antarctic species (0.094–0.17), and more than 10 times higher that in L. mormyrus (0.028) and S. cabrilla (0.018). The mean free spacing of mitochondria decreased from 9.5 μm in S. cabrilla to 3.1 μm in Champsocephalus esox (Table 1). The three-way ANOVA revealed highly significant main effects of activity pattern (F(1,178)=42.2; P<0.0001) and species (geographical zone) (F(5,178)=11.0; P<0.0001) on mitochondrial free spacing.
The volume density of myofibrils, Vv(my,f), showed considerable inter-specific variation and was approximately 0.4 in T. newnesi, L. nudifrons, C. esox, E. maclovinus, P. magellanica and P. tessellata, 0.52 in Coris julius and highest in L. mormyrus (0.74) and S. cabrilla (0.80). There were significant effects of activity (F(1,178)=18.5; P<0.001), geographical zone (F(2,178)=8.29; P<0.05) and species (geographical zone) (F(5,178)=15.0; P<0.001) on the volume density of myofibrils. A Fisher’s protected least significant difference test revealed that the Mediterranean species have a significantly higher volume density of myofibrils than the Antarctic and sub-Antarctic species (P<0.05). The spatial organization of myofibrils varied considerably among the species. Champsocephalus esox showed the most distinctive organisation, with single myofibrils surrounded by mitochondria (Figs 1C, 4A). In muscle fibres of the other species, myofibrils were arranged in ribbons of which the thickness varied among the biogeographical zones (Fig. 1). In the Antarctic species, the ribbons were one fibril thick (Fig. 1A,B), whereas in the sub-Antarctic species they formed thicker clusters (Figs 1D,E, 2A). In the Mediterranean species, the ribbons of myofibrils were more densely packed (Fig. 2B–D) with those closest to the sarcolemma typically flattened in cross-section, resembling the spokes in a bicycle wheel (Fig. 2D). The volume density of intracellular lipid droplets was not correlated with biogeographical zone or locomotory habit and the maximum value was 0.03 (Table 1). The volume densities of mitochondria and myofibrils in the red muscle fibres of E. maclovinus were similar in summer- and winter-caught fish (Table 1). However, the surface density of mitochondrial clusters was significantly higher (P<0.01) and the mean free spacing of mitochondrial clusters was significantly lower (P<0.01) in summer than winter fish (one-way ANOVA) (Table 1).
Mitochondrial cristae surface densities were similar among the species examined in this study, ranging from 36 to 50 μm2μm−3 (Table 2, Fig. 4A–C). Mitochondria from the sub-Antarctic ice-fish Champsocephalus esox showed a considerably higher cristae surface density (43.9 μm2μm−3, Fig. 4B) than previously found for two species of Antarctic icefish (25–28μm2μm−3) (Archer and Johnston, 1991; Table 2). The surface area of mitochondrial cristae per unit fibre volume was approximately twice as high in the pelagic sub-Antarctic channichthyid Champsocephalus esox as in either Antarctic icefishes or pelagic red-blooded sub-Antarctic species (Table 2). Among the Mediterranean species studied, the surface area of mitochondrial cristae per unit fibre volume was three times higher in Coris julius than in Lithognathus mormyrus (Table 2).
Mitochondrial respiration
The state 3 respiration rates of mitochondria with pyruvate as substrate are shown in Fig. 5. RCR values ranged from 4.2 to 10.0 for preparations from Antarctic species, from 4.0 to 7.7 for the sub-Antarctic species and were approximately 5 for S. cabrilla, showing that the mitochondria were coupled. The rate of mitochondrial oxygen consumption increased with temperature for each species with a Q10 of approximately 2.0. Measured at their respective habitat temperatures, the maximal rate of oxygen consumption was twice a high in pelagic and semi-pelagic sub-Antarctic as in Antarctic species. State 3 respiration rates of isolated mitochondria from Lepidonotothen nudifrons and T. newnesi at −1°C were only 37.5% of that for S. cabrilla at 20°C (Fig. 5). The combined data for these perciform species could be best described by a single quadratic relationship:
DISCUSSION
To a first approximation, in the perciform species studied, the maximum rate of oxygen consumption per milligram of isolated red muscle mitochondrial protein was a direct function of assay temperature (Fig. 5). Pyruvate, the carbon substrate that these mitochondria were oxidizing, generally yields maximum rates of state 3 respiration in fish mitochondria (Guderley and St Pierre, 1996). Similar results, although with more variance in the data, have been reported for mitochondria isolated from several perciform, salmoniform and cypriniform familes (Johnston et al. 1994; Guderley and St Pierre, 1996). In the present study, the oxidative capacities of red muscle mitochondria isolated from the Antarctic species were either within or slightly below the 95% confidence limits of the relationship between assay temperature and mitochondrial oxygen uptake for the overall data set. The predicted rate of oxygen consumption at 1°C was slightly higher if the Antarctic species were excluded from the regression. The results therefore clearly indicate that little significant up-regulation of maximum oxidative capacity per milligram mitochondrial protein has occurred in the Antarctic perciforms.
Although insufficient information is available to make a quantitative phylogenetic analysis, it is interesting to note that the two sub-Antarctic species have similar mitochondrial respiration rates and yet the dorado (Patagonotothen magellanica) is thought to be more distantly related to the róbalo, Eleginops maclovinus than to the Antarctic Nototheniidae Trematomus newnesi and Lepidonothen nudifrons. The oxidative capacities of mitochondria from T. newnesi and Lepidonotothen nudifrons were significantly higher than we reported previously for mitochondria from another Antarctic species Notothenia coriiceps (P<0.05; Mann–Whitney U-test) (Fig. 5; Johnston et al. 1994). The available information on the natural history, morphology and behaviour of these fish suggests that N. coriiceps is much less active than the other species (Eastman, 1993; North, 1996). Field observations of an individual N. coriiceps with an underwater television camera showed that it stayed within 3 m of a small cave for over 6 days, spending only 1.2% of the time swimming, usually for short periods at less than 2 body lengths s−1 (North, 1996). The lower oxidative capacity of the isolated mitochondria of this species may be related to this more sedentary behaviour.
An enhanced mitochondrial cristae surface density, at constant mitochondrial densities, could expand tissue aerobic capacity in the absence of increases in oxidative capacity per milligram mitochondrial protein. However, as with the latter, cristae density showed no evidence of up-regulation in Antarctic species (Table 2). Interestingly, the mitochondrial cristae surface density for the pelagic sub-Antarctic icefish Champsocephalus esox was considerably higher than that of its Antarctic counterparts, as was the cristae surface area per unit fibre volume (Table 2). The cristae surface densities we report here are similar to or higher than values reported for mitochondria from a wide variety of mammalian skeletal muscle (20–40 μm2μm−3) (Hoppeler and Lindstedt, 1985; Schwerzmann et al. 1989), and mammalian and reptilian heart muscle (35–60 μm2μm−3) (Else and Hulbert, 1985), but are lower than those reported for hummingbird flight muscle (58 μm2μm−3) (Suarez et al. 1991) or tuna red muscle (63–70 μm2μm−3) (Moyes et al. 1992). Srere (1985) calculated that, in order to have sufficient matrix space for two average-sized Krebs cycle enzymes, the maximum possible cristae surface density would be approximately 83 μm2μm−3. The cristae surface densities of mitochondria from hummingbird flight muscle and tuna red myotomal muscle approach this value.
Mitochondrial cristae surface density is not a particularly good predictor of mitochondrial oxidative capacity corrected for the influence of temperature. Within the perciforms, the oxidative capacities of isolated mitochondria from E. maclovinus and Paranotothenia magellanica were identical, whereas their mitochondrial cristae surface densities differed (39 versus 45; Fig. 5; Table 2). Whereas mitochondria from tuna muscle possess cristae surface densities of 63–70 μm2μm−3, their oxidative capacities at 25°C (106 nmol O mg−1 mitochondrial protein min−1) (Moyes et al. 1992) are no higher than the rates expected at 25°C for the perciform mitochondria we studied (Fig. 5), for which cristae surface densities ranged from 36 to 50 μm2μm−3. These data suggest that, if the packing limits of cristae and matrix proteins are respected, oxidation rates per milligram mitochondrial protein at a common temperature change little. Further, these interspecific comparisons suggest a fairly constant protein content of the mitochondrial inner membrane. Nonetheless, the fourfold higher oxidation rates found at 25°C for tuna heart mitochondria (400 nmol O mg−1 mitochondrial protein min−1), which have a cristae surface density of 56 μm2μm−3 (Moyes et al. 1992), suggest that oxidative capacity may be increased by other mechanisms, perhaps membrane lipid composition or specific activities of mitochondrial proteins.
The mitochondrial volume density of red muscle fibres was significantly higher in the pelagic haemoglobin-less icefish Champsocephalus esox than in any of the red-blooded fish (Table 1). The contractile filaments in this species were arranged as individual myofibrils or as small clusters of myofibrils entirely surrounded by mitochondria, accounting for approximately 50% of muscle fibre volume, as was previously reported for the Antarctic icefishes Chaenocephalus aceratus (Johnston, 1987) and Champsocephalus gunnari (Archer and Johnston, 1991). Interestingly, the cristae surface density of mitochondria (43.9 μm2μm−3) in Champsocephalus esox is comparable with that of the red-blooded species (Table 2) and was considerably higher than those reported for Chaenocephalus aceratus (28.2 μm2μm−3) and Champsocephalus gunnari (25.2 μm2μm−3) (Archer and Johnston, 1991). Mitochondrial packing in icefishes may approach the theoretical maximum for a locomotory muscle beyond which there is insufficient space to accommodate sufficient myofibrils to generate the required stress. The maximum stress exerted by the myofibrils is thought to be relatively constant between species with different body temperatures (Johnston, 1985). Higher values for Vv(mt,f) of 0.63 have been reported in the swordfish Xiphus gladius for the heater organ, which is formed from a highly modified extraoccular muscle (Block, 1994). However, in this case, myofibrils are absent, although the sarcoplasmic reticulum is more extensive than in any fish locomotory muscle, a feature thought to be related to the futile cycling of Ca2+ to activate thermogenesis (Block, 1994).
Among the relatively active pelagic or semi-pelagic red-blooded species studied (Figs 1, 2), the volume density of mitochondria was between 0.27 and 0.33 (Table 1). Examination of pectoral fin adductor muscles in a wider range of Perciformes that use a predominantly labriform style of swimming suggests that there is no significant relationship between Vv(mt,f) and habitat temperature (Fig. 6). Although this conclusion relies heavily on the data point for Coris julius (point 10 in Fig. 6), it is supported by measurements of mitochondrial volume density in red myotomal fibres from other active pelagic species, e.g. 0.46 in the European anchovy (Engraulis encrasicolus), a filter feeder, living at 18–25°C (Johnston, 1982b) and 0.29 in the skipjack tuna (Katsuwonus pelamis) (Mathieu-Costello et al. 1992; Moyes et al. 1992). The latter is a tropical species exhibiting regional endothermy and it can maintain elevated red muscle temperatures of up to 31°C during bouts of active swimming (Stevens and Neill, 1978).
In contrast, among the demersal species and moderately active mid-water fish, there was a highly significant inverse correlation between mitochondrial volume density and the habitat temperature at which the fish were captured (adjusted r2=0.64, P=0.03) (Fig. 6). It has been suggested that the proliferation of mitochondria with cold-acclimation represents a mechanism to compensate for the reduction in the diffusion coefficients of cytostolic metabolites by decreasing the mean free spacing between mitochondria (Tyler and Sidell, 1994). However, Hubley et al. (1997) recently utilised pulsed-field gradient 31P nuclear magnetic resonance spectroscopy and a mathematical reaction–diffusion model to calculate profiles of phosphocreatine and ADP concentration in muscle fibres of goldfish Carassius auratus acclimated to either 5°C or 25°C. Their results showed constant metabolite gradients within the red muscle fibres over time regardless of temperature, thermal acclimation state or mitochondrial volume density. The results clearly indicate that the proximal stimulus for temperature-induced changes in mitochondrial volume density is not a disruption of intracellular diffusion of high-energy phosphates. Nonetheless, limitations in the diffusion of other metabolites, specifically carbon substrates, may provide such a stimulus. Mean free spacing of mitochondria is lower in active pelagic fish and at high temperatures; however, this may well represent a secondary consequence rather than a primary aim of the increase in mitochondrial volume density.
The total aerobic capacity is likely to be some function of the number and functional capacity of the mitochondria. However, quantitative predictions concerning the maximal rate of oxygen consumption in vivo are difficult since this will depend among other things on intracellular pH (Moyes et al. 1988), the availability of ADP and NADH (Brand and Murphy, 1987) and the delivery of oxygen and fuel by the capillary circulation (Mathieu-Costello et al. 1992). Nevertheless, for active pelagic species, our results indicate that the maximum aerobic capacity of red muscle is substantially lower in polar than in tropical fish. From the regression line in Fig. 6, the volume density of mitochondria in the red muscle of relatively inactive fish decreased approximately twofold between −1 and 22.°C, whereas the maximal (state 3) rate of oxygen uptake of the isolated mitochondria increased almost fourfold (Fig. 5). Thus, for less-active swimmers, the proliferation of mitochondria in polar relative to Mediterranean species would serve to compensate partially for the reduced mitochondrial oxidative capacity per milligram protein at low temperatures.
There is evidence that the factorial aerobic scope for activity is approximately 5–7 for non-athletic fish species (Forster et al. 1987; Gordon et al. 1989; Johnston et al. 1991). In a study of sedentary demersal fish with similar lifestyles, it was found that, after correcting for differences in body mass, resting metabolic rate was three- to fourfold lower in Antarctic than in Indo-West Pacific species (Johnston et al. 1991). If a fairly constant relationship between resting and maximum aerobic metabolism (Wieser, 1985) pertains to these species, a low resting rate suggests a low maximal rate. The results of our current study are therefore consistent with the relatively low aerobic scope for activity that has been measured in polar species (Forster et al. 1987; Johnston et al. 1991; Bushnell et al. 1994). Although exploitation of cold seas is associated with extensive modifications in membrane lipid composition (Hazel and Williams, 1990) and the ligand-binding properties of enzymes (Somero et al. 1996), and such modifications are certain to have been required to permit mitochondrial activity at sub-zero temperatures, maximum rates of mitochondrial substrate oxidation (per milligram mitochondrial protein) have not been up-regulated to attain values similar to those found in species specialised for life in warm seas.
ACKNOWLEDGEMENTS
This work was supported by a grant from the International Scientific Co-operation Program of the European Union and by the Natural Environment Research Council of the UK. We are also grateful to Marguerite Abercromby and Irvine Davidson for skilled technical support and to the British Antarctic Survey for supplying the live Antarctic fish.