Brain and Body Oxygen Requirements of Gnathonemus Petersii, a Fish with an Exceptionally Large Brain

The brain of man has hitherto been thought to account for the largest fraction of whole-body energy expenditure in vertebrates. The human brain consumes some 20 % of the O₂ taken up by the resting body (Crile, 1941; Mink et al. 1981). This has been put forward as the only gross physiological measure that makes the human brain unique among vertebrates, since its relative size per se, although corresponding to an impressive 2 % of body mass, is surpassed by that of several small mammals and birds (Crile and Quiring, 1940; Krompecher and Lipák, 1966). It has been pointed out that other vertebrates are remarkably uniform with respect to the ratio of brain to body O₂ consumption, which is of the order of 0.02–0.08, essentially regardless of body size (Crile, 1941; Mink et al. 1981). In fact, the only vertebrates known to possess brains consuming more than 10 % of total O₂ intake are some primates (11–13 %) and man (Mink et al. 1981). It is possible that the brain of the porpoise Phocaena phocaena also consumes a comparatively high fraction of the total O₂ inspired, since it constitutes 1.2 % of body mass (Crile and Quiring, 1940).

The uniqueness of the human brain with regard to its exceptionally high rate of energy use appears to have remained undisputed. However, in Africa, there is a group of weakly electric fishes, the elephant nose fishes (Mormyridae), that has been noted for having unusually large brains (Erdl, 1846), a character that is probably, at least in part, related to their ability to communicate by generating and perceiving electric fields (Hopkins, 1983; Nieuwenhuys and Nicholson, 1969). In contrast to man, it is the cerebellum, and not the telencephalon (cerebrum) that is greatly enlarged in these fishes, the structure being appropriately named the gigantocerebellum. From an evolutionary and ecophysiological point of view, the energetic cost of this brain may be more important than its size per se. As ectothermic vertebrates, these fish are expected to have low rates of whole-body O₂ consumption which could result in the brain comprising a relatively large fraction of the whole-body energy budget. However, brain and body O₂ consumption have never been measured in mormyrids. Moreover, since the vertebrate brain is normally very sensitive to low ⁠, the hypoxia-tolerance of an animal with such a huge brain is of particular interest.

The aim of the present study was to measure brain and body O₂ consumption in the west African mormyrid Gnathonemus petersii, so that the ratio of brain to body O₂ consumption could be estimated. In addition, the ability of this fish to regulate O₂ consumption and electric pulse rate in response to falling O₂ levels, as well as its tolerance to severe hypoxia, were examined.

Gnathonemus petersii Günther were obtained from a fish dealer in Uppsala, Sweden. They were held in an aquarium at 26 °C (16 fish in 250 l) and fed chironomid larvae daily until satiation, for at least 1 month before the experiments. All experiments were carried out at 26 °C.

Four experiments were performed on the 16 fish available. In the first experiment, body O₂ consumption was determined in four fish during decreasing O₂ levels in a closed respirometer. In the second experiment, the brains of four fish were removed and O₂ consumption of brain slices was determined. In the third experiment, five fish were exposed to more rapidly falling O₂ levels by bubbling the water with N₂ and their behaviour was observed. In a final group of three fish, body O₂ consumption and the discharge rate of the electric organ were determined simultaneously during falling O₂ levels in a closed respirometer. Body O₂ consumption rates of G. petersii at different ambient O₂ concentrations were measured using closed respirometry in a 2 l respirometer as previously described (Nilsson, 1992). Each fish, which had not been fed for 24 h, was allowed to respire in the respirometer until the O₂ concentration reached approximately 0.3 mg l^-1, which took 780–1500 min (depending on fish size). The rate of O₂ consumption was calculated over 6 min intervals. After having been put into the respirometer, the fish initially showed a relatively high level of locomotor activity, causing rates of O₂ consumption to increase above resting values. In consequence, the first 200 min of the experiments were excluded from the calculations. The critical ambient O₂ level was determined as follows. A linear regression line was calculated from the O₂ consumption values measured at ambient O₂ levels above 1.5 mg l^-1 (since no tendency for a break in the O₂ consumption curve was seen above that level; Fig. 1). The critical ambient O₂ level was assumed to correspond to the first data point that fell below this line if all subsequent values of O₂ consumption were also below the regression line.

The electric discharge of the fish was recorded using two stainless-steel rods (30 mm long, 3 mm diameter) introduced into the respirometer and connected to an a.c. amplifier. The discharge frequency (Fig. 2) was calculated from the time taken to record 4096 discharges. The frequencies recorded (generally 10–15 pulses s^-1) constituted average values over 4.5–7 min periods.

Brain slices (300 μm thick) were prepared from the whole lateral half of a brain using a Vibratome (Vibroslice 752; WPI, UK). Their O₂ consumption rates were determined (within 1 h of preparation) in a 1.6 ml thermostatted (26 °C) glass chamber equipped with a Clark O₂ electrode. Approximately 40 mg of randomly selected brain slices were used for each measurement. The slice preparation and O₂ consumption measurements were carried out in oxygenated brain-slice Ringer’s solution prepared as described previously (Johansson et al. 1995).

Values are presented as means ± S.D.

Table 1 shows that the brain of G. petersii constitutes 3.1 % of the body mass and that it is responsible for 60 % of whole-body O₂ consumption. The corresponding values for the human brain are approximately 2.3 % and 20 %, respectively, while those for other vertebrates are considerably lower. The estimated in situ values for O₂ consumption of the G. petersii brain and the brains of other fishes were obtained by multiplying brain-slice O₂ consumption by a factor of two (see Discussion).

Whole-body O₂ consumption rates measured during decreasing ambient (Figs 1, 2) showed that the critical O₂ level for steady rates of O₂ consumption was 0.77±0.10 mg O₂ l^-1 (N=7), i.e. an O₂ saturation of approximately 10 % at 26 °C. At a normoxic O₂ level in water (5 mg l^-1), the rate of O₂ consumption of the fish was 0.102±0.020 mg g^-1 h^-1 (Table 1; Figs 1, 2).

In a separate experiment, five fish were put in a 5 l aquarium in which the O₂ content was lowered to less than 0.3 mg l^-1 within 2 h by bubbling with N₂. When the O₂ concentration fell below 0.8 mg l^-1, the fish started to gulp air (in this case N₂) from the surface, clearly an attempt to increase their O₂ uptake. Finally, when the O₂ content fell below 0.3 mg l^-1, the experiment had to be abandoned, since within 14–18 min all fish had problems with maintaining an upright body posture, clearly indicating an inability to tolerate such a low ambient O₂ concentration.

Fig. 2 shows electric discharge frequencies and O₂ consumption rates of three individuals in response to falling ambient O₂ levels. Both the electric activity and the O₂ consumption remained relatively constant down to the same critical O₂ level (0.82±0.05 mg l^-1 and 0.77±0.1 mg l^-1, respectively).

The brain of the mormyrid fish G. petersii consumes a larger fraction, i.e. approximately 60 %, of the whole-body O₂ intake than any other vertebrate brain studied, including the human brain, which had been thought to hold the record in this respect. The in situ O₂ consumption of the G. petersii brain (Table 1) was estimated by multiplying the in vitro values of brain slices by a factor of two. This was done because brain slices are energetically depressed, showing very low spontaneous electric activity. Brain slices from mammals consume only about half of the O₂ consumed by the intact brain (McIlwain and Bachelard, 1971; Mink et al. 1981; Lipton and Whittingham, 1984). The small size of G. petersii did not allow the catheterization of blood vessels necessary for measuring O₂ consumption of the brain in situ.

Whereas total energy expenditures of ectothermic vertebrates (Fig. 3B) are about 1/13 of those of endotherms at the same temperature and body size, mass-specific energy expenditures of the brains of ectothermic and endothermic vertebrates are much more similar (Fig. 3A). The average brain O₂ consumption rate of the three ectothermic vertebrates shown in Fig. 3A (derived from Table 1) is 4.95 mg g^-1 h^-1 (at 37 °C), which is not much lower than the 7.74 mg g^-1 h^-1 expected for an endothermic brain of the same average size (deduced from the regression line for endothermic brains in Fig. 3A). In fact, the O₂ consumption rate of the rainbow trout brain lies on the regression line for endothermic brains. Thus, with regard to the whole-body energy budget, it is comparatively more expensive for an ectothermic vertebrate to have a large brain. This may be a reason why most ectothermic vertebrates have relatively small brains, resulting in the strikingly constant ratio of brain to body O₂ consumption found for both ectothermic and endothermic vertebrates. Consequently, the fact that G. petersii is an ectothermic vertebrate and has such a huge brain makes this organ an exceptionally expensive part of the whole body.

As mentioned above, there are several small endothermic vertebrates, notably bats, mice, swallows, crows and sparrows, that have brains which make up 2.6–3.7 % of the body mass (Crile and Quiring, 1940; Mink et al. 1981). However, the very high whole-body metabolic rates of these small endotherms make their brains comparatively less costly (using only 4–8 % of the body energy budget), because in vertebrates the scaling factor of the mass-specific rate of energy expenditure is much lower for the brain than for the whole body (Fig. 3; Mink et al. 1981; Schmidt-Nielsen, 1984).

The Q₁₀ for the rate of O₂ consumption of brain tissue is about 2.1 in vertebrates (Mink et al. 1981), suggesting that the specific rate of O₂ consumption of the G. petersii brain at 37 °C would be 4.57 mg g^-1 h^-1. This is similar to the value of 6.02 mg g^-1 h^-1 given for the rat brain by Nilsson and Siesjö (1976) and it is higher than the oxygen consumption of the human brain (2.61 mg g^-1 h^-1). The low value of the latter reflects the slightly negative scaling exponent of the metabolic rate of the brain. Among fishes, the brain of G. petersii does not have a uniquely high mass-specific rate of O₂ consumption; the value reported in this study was lower than the in situ value of 2.20 mg g^-1 h^-1 at 20 °C estimated from previous brain slice measurements in salmonids (Table 1).

The reason why ectothermic brains do not differ much from endothermic brains with regard to the rate of energy utilisation is probably related to the fact that ionic leakage, which occurs at much higher rates and is largely coupled to heat production in endothermic vertebrates (Hulbert and Else, 1990), plays only a minor role in the energy expenditure of the brain. Ion movements involved in electric and synaptic activity are responsible for the largest portion of the energy used by the brain (Erecinska and Silver, 1989), and in this respect there is probably little difference between endothermic and ectothermic brains. Consequently, when exposed to anoxia, rainbow trout brain ion homeostasis becomes disrupted due to energy failure as fast as does that of the mammalian brain (Nilsson et al. 1993).

The striking hypoxia tolerance of G. petersii may appear paradoxical for an animal with such an energetically costly brain. G. petersii has the ability to regulate the rate of O₂ uptake at low ⁠, displaying a critical value for O₂ consumption at approximately 10 % O₂ saturation. However, the capacity to utilise low ambient levels of O₂ is probably important for protecting the brain from hypoxic damage. Mormyrids inhabit turbid eutrophic tropical waters in which hypoxia is likely to be a common event. In most vertebrates, the brain is strongly dependent on aerobic energy production, being unable to meet its high ATP demand by anaerobic glycolysis alone (Lutz and Nilsson, 1994). In G. petersii, the brain is likely to be the first organ to be affected by energy failure during anoxia; the results presented in Table 1 suggest that its brain consumes O₂ at 20 times the rate of the average body tissue. Indeed, bubbling the water with N₂ revealed that G. petersii cannot tolerate an ambient O₂ concentration below 0.3 mg l^-1, suggesting that its capacity to produce ATP anaerobically is not sufficient to match its rate of ATP consumption. In nature, it is likely that G. petersii relies on gulping air or oxygenated surface water when the ambient O₂ concentration falls below the critical level. Aquatic surface respiration in response to low ambient is displayed by numerous fish species, especially in the tropics (Kramer and McClure, 1982).

The discharge frequency of the electric organ of G. petersii was maintained at a relatively constant rate down to approximately the same critical ambient as for O₂ consumption (Fig. 2). This indicates that G. petersii depresses neither its metabolic rate nor its nervous activity in response to progressive hypoxia, the discharge rate of the electric organ being under central nervous control (Dye and Meyer, 1986). As in other mormyrid fishes, the electric organ of G. petersii delivers electric pulses with a constant amplitude, the electric activity being modulated by changing the discharge rate (Teyssedre and Serrier, 1986). Although not measured in this study, it seems unlikely that this species is able to lower its pulse amplitude in order to save energy in response to progressive hypoxia.

This study reveals that the limit on the proportion of the total energy budget that can be devoted to central nervous activity is much higher than previously known. Since mormyrids seem to be successful fishes, making up some 10 % of the African species of freshwater fish (Roberts, 1975), and are abundant enough to be of local commercial importance (Petr, 1968), it is relevant to ask what advantageous abilities are provided by this exceptionally costly brain. It is likely that these abilities not only include electroreception, a highly effective means of orientation and communication in turbid waters, but also social communication, which appears to be well developed in mormyrids (Kramer, 1990; Moller et al. 1989). Interestingly, the weakly electric fish of the group Gymnotoidei of South America, which seem to have a similar electric sensory system (Hopkins, 1983), do not show the extreme increase in brain size displayed by the Mormyridae (Nieuwenhuys and Nicholson, 1969).

Finally, from the present results and from other data discussed above, it is clear that the human brain is not unique in gross physiological terms, neither with regard to its relative or absolute size, nor with regard to its mass-specific or fractional metabolic rate.

I thank Mr Sven Olof Carlsson for valuable technical assistance. This study was financially supported by the Swedish Council for Forestry and Agricultural Research, the Swedish Natural Science Research Council, and the Magn. Bergvall Foundation.