Metabolism of the Perfused Swtmbladder of the European EEL: Oxygen, Carbon Dioxide, Glucose and Lactate Balance

Many teleost fishes have developed a gas-filled swimbladder which enables them to keep neutral buoyancy at varying water depths. It is generally accepted that the gas enters the bladder from the blood by diffusion, and it is assumed that the high gas partial pressures in the blood perfusing the swimbladder are generated by acidification of the blood in the gas gland in conjunction with the countercurrent gas exchange in the rete mirabile (Steen, 1970; Fänge, 1983). In fact, the gas gland is known to produce lactic acid even at high O₂ levels. This acid enhances, via the Bohr and Root effects, the O₂ partial pressure and, by the conversion of HCO₃⁻ to CO₂, the CO₂ partial pressure. The formation of lactate by gas gland cells has been demonstrated both in tissue preparations (Ball et al. 1955; D’Aoust, 1970; Deck, 1970) and in vivo (Kuhn et al. 1962; Steen, 1963b; Enns et al. 1967). In this model, CO₂ in the swimbladder would derive either from oxidative metabolism or from HCO₃⁻.

Oxidative metabolism has been estimated to contribute only 5 % of the total CO₂ formed (Wittenberg et al. 1964). Ball et al. (1955), Wittenberg et al. (1964) and D’Aoust (1970) found that most, if not all the CO₂ was liberated from bicarbonate, presumably by lactic acid. This assumption neglects, however, the presence of nonbicarbonate buffers, which would buffer some of the H⁺ released from lactic acid, thus reducing the amount of H⁺ available for conversion of HCO₃⁻ to CO₂. Therefore, a lactate/CO₂ ratio of unity, found by Ball et al. (1955) and D’Aoust (1970), can only be explained if CO₂ derives from sources other than bicarbonate.

Such a source could be the decarboxylation reaction of the pentose phosphate shunt, which yields CO₂ from carbohydrates without consuming O₂. Bostrom et al. (1972) measured high activities of enzymes characteristic of the pentose phosphate shunt in the gas gland tissue of the cod, and increases in CO₂ content in blood passing through the gas gland have indeed been measured directly (Steen, 1963b; Pelster et al. 1988b). The contribution of CO₂via this route has not been estimated yet.

In the present study, we have measured the metabolic activity in the salineperfused eel swimbladder to evaluate the contributions of the various pathways to CO₂ production. The saline-perfused preparation was preferred to the more physiological, blood-perfused swimbladder, as it more easily allows measurement of the perfusion rate and quantitative estimation of metabolic rates.

Specimens of the freshwater-adapted European eel (Anguilla anguilla; body mass 500–1100 g) were purchased from a local supplier and kept in a freshwater aquarium at 12–14°C until the experiment.

The animals were anaesthetized by adding MS 222 (0·1 g l⁻¹) to the water and were then placed into an ‘eel holder’, similar to that used by Steen (1963a). The gills were irrigated with well-oxygenated tap water (flow rate 4·51 min⁻¹), containing MS 222 at a concentration of approximately 0·05 g l⁻¹, a level sufficient to keep the animal anaesthetized with stable ventilatory activity.

Preparation of animals was essentially the same as described elsewhere (H. Kobayashi, B. Pelster & P. Scheid, in preparation). Briefly, the body wall was opened ventrally, the swimbladder was exposed and carefully freed from connective tissue. The duct connecting the absorption and secretion part of the bladder was tied off, and small blood vessels and anastomoses bypassing the rete mirabile were carefully ligated without damaging rete vessels or the swimbladder wall. The gas secretion rate was too low to be measured accurately by a catheter inserted into the swimbladder (less than 0–1ml h⁻¹). All experiments were carried out at 20 –22°C.

The artery and vein at the heart pole (ai and ve in Fig. 1) were occlusively cannulated using PE 50 and PE 100 catheters, respectively, and the rete vessels were perfused with saline solution at 0·13–0·40mlmin⁻¹ using an infusion pump (Precidor, Type 5003, Infors AG, Basel, Switzerland). This solution was a heparinized (100 i.u. ml⁻¹) Ringer’s solution which contained (in mmol l⁻¹) NaCl, 129; KC1, 5; MgSO₄, 0·9; CaCl₂, 1·1; glucose, 15; insulin, 8i.u. The colloid osmotic pressure was adjusted to that of eel plasma by adding 5 g 1⁻¹ Dextran FP 70 (Serva, Heidelberg, FRG). The saline contained no bicarbonate and was equilibrated with air so as to remove nearly all CO₂.

Samples of the perfusate were collected at the venous outlet (ve), and their composition was compared with that of the perfusate at the inflow (ai). Venous collection was about every 30 min for 3–4 h. For calculating metabolic rates from perfusion rate and concentrations in inflow and outflow, steady-state conditions have to prevail. Since countercurrent exchange in the rete during normal flow conditions (perfusion type C) retards attainment of steady state, a serial type of rete flow (type S) was achieved in some experiments by occluding the afferent artery of one and the efferent vein of the other rete at the heart pole (Fig. 1), this flow type enhanced attainment of steady state. Measurements were made in these experiments during both type C and type S flow.

Saline samples were analysed for oxygen content (⁠⁠; using the method of Tucker, 1967) and total CO₂ content (⁠⁠; using the method of Cameron, 1971) immediately after collection. For analysis of glucose and lactate concentrations, part of the sample was deproteinized with perchloric acid and neutralized with K₂CO₃ or KOH. Employing this procedure, no extrusion of proteins from the tissue into the perfusate was detected. The assays were performed enzymatically as outlined by Bergmeyer (1974).

The rates of production of CO₂ and lactate (Ṁ_La) and the rates of consumption of O₂ and glucose (Ṁ_Glu) were calculated by mass balance from the perfusion rate (Q̇) and the concentration difference between inflow and outflow. Student’s t-tests were used to test for statistical significance, and P< 0·01 was accepted as the limit of significance.

At the onset of perfusion, there were high concentrations of lactate and CO₂ in the venous outflow (ve) which gradually decreased to a steady-state level. For both substances this was higher than the arterial level (ai; Fig. 2). For calculation of metabolic rates only steady-state values were taken into account.

Table 1 shows concentrations of O₂, CO₂ and lactate in inflow (ai) and outflow (ve) of the rete during perfusion at two flow rates when there was no countercurrent flow (type S). The observed differences in these concentrations between ai and ve were clearly dependent on the flow rate. Doubling the flow rate resulted in a reduction of the (ve–ai) differences by a factor of about 2.

The rates of O₂ consumption and CO₂ production ⁠, calculated from concentration differences in individual experiments, are listed in Table 2. There were no differences between the flow types. In each experiment the CO₂ production rate by far exceeded the O₂ consumption rate. The resulting values of the respiratory exchange ratio, ⁠, are significantly above unity. The rates of glucose uptake (Ṁ_Glu) and lactate release (Ṁ_La) are presented in Table 3. On average, the rate of glucose consumption was higher than the rate of lactate production.

For a quantitative evaluation of the metabolic activities of the gas gland its perfusion rate had to be known. Since blood flow rate cannot easily be obtained in the eel swimbladder without altering its function, we had to apply the technique of saline perfusion of the vascularly isolated tissue. The CO₂ and lactate values measured in this study, however, are similar to values that we obtained earlier by non-obstructive blood collection in intact eel swimbladder (Pelster et al. 1988b). We, therefore, assume the values measured in this study to be representative for eel swimbladder under physiological conditions.

The important results of this study are the demonstration of the high rate of CO2 production in the swimbladder tissue and the high respiratory exchange ratio, which significantly exceeded the value of unity expected for oxidative glucose breakdown. The values were calculated assuming no gas secretion into the swimbladder, which was indeed below the detection level. In our experiments the CO₂ may derive from three sources: washout from the tissue, oxidative metabolism and anaerobic decarboxylation reactions.

As the CO₂ content in the perfusing solution was kept low, to ensure a high sensitivity for arteriovenous CO₂ differences, the swimbladder tissue had probably been cleansed of most of the CO2 and HCO₃⁻ during the perfusion period before measurements were performed. An enhanced conversion of HCO₃⁻ to CO₂ is to be expected as a result of lactic acid formation.

The exact amount of HCO₃⁻ /CO₂ washed out from the tissue is difficult to estimate. Let us consider first the conversion of HCO₃⁻ to CO₂ in the tissue, effected by the protons liberated from lactic acid, and its subsequent release into the perfusate. The amount of lactate produced (Table 1) would, indeed, suggest a large contribution to the CO₂ release into the perfusate, but there are two other possibilities for the H⁺ apart from combining with HCO₃⁻. One is binding to nonbicarbonate buffers in the tissue, the other binding to any buffer in the perfusate and washout with it. The nominal buffer value of the perfusate was very low, and the pH change in the perfusate from ai to ve of 1–2 units suggests that the amount of H⁺ washed out is probably low. We do not know the buffer value of the swimbladder tissue and, thus, cannot estimate the amount of CO₂ washed out from it by considering the H⁺ liberated from lactic acid.

There is another reason to suggest that the contribution of HCO₃⁻ /CO₂ washout from tissue was probably small. Assuming a bicarbonate concentration of approximately 2 mmol l⁻¹, typical of muscle tissue, the tissue mass of the swimbladder (about 0·4g for an 800 g eel; Ball et al. 1955; Wittenberg et al. 1964) would contain less than l μmol of HCO₃⁻. Under steady-state conditions (see Fig. 2) this cannot contribute very much to the observed rate of CO₂ formation.

The rate of CO₂ production by glucose oxidation can best be estimated in our experiments from the O₂ consumption rate, which suggests that 12 nmol min⁻¹ or about 20 % of the CO₂ derives from this pathway. It thus appears that the major fraction of th.e CO₂ formed must stem from the third source.

An example of anaerobic CO₂ formation is the pentose phosphate shunt. This pathway allows formation of CO₂ by decarboxylation of 6-phosphogluconate formed enzymatically from glucose. The pentose phosphate shunt is known to occur in ocular tissues of fish, where it contributes significantly to the glucose metabolism (Hoffert & Fromm, 1970). Boström et al. (1972) found that in the cod glucose-6-phosphate dehydrogenase and 6-phosphogluconate dehydrogenase, key enzymes of the pentose phosphate shunt, displayed higher activities in gas gland cells than in muscle cells. In preliminary measurements we found activities of these two enzymes in the eel gas gland similar in magnitude to those in liver tissue and about 10 times higher than their activities in muscle tissue. It is, thus, to be expected that part of the glucose should be diverted to the pentose phosphate shunt to form CO₂ by decarboxylation catalysed by 6-phosphogluconate dehydrogenase. In this case, an enzymatic reaction reoxidizing the NADPH produced in the pentose phosphate shunt is necessary. This reoxidation might occur by chain elongation of fatty acids.

Another possibility for anaerobic CO₂ formation has been described for goldfish muscle (Van den Thillart & Verbeek, 1982; Mourik, 1982). This CO₂ formation, which is combined with ethanol formation (Shoubridge & Hochachka, 1980), is predominantly located in the mitochondria of red muscle cells. The observed activities of the pentose phosphate shunt enzymes, and the histological observation that gas gland tissue contains only few mitochondria (Dorn, 1961), lead us to favour the pentose phosphate shunt as a possible pathway for anaerobic CO₂ formation.

Since the pentose phosphate shunt requires glucose, more glucose should be degraded in our experiments than would be expected from lactate formation and oxidative metabolism. The balance of glucose metabolism is, therefore, of interest in our study.

The glycogen stores in swimbladder tissue are too small to fuel glycolysis for more than a few minutes (D’Aoust, 1970), and blood (or saline) glucose is expected to be the main source for metabolism. Three pathways of glucose degradation have to be considered in our study: aerobic oxidation to CO₂, anaerobic fermentation to lactic acid and anaerobic decarboxylation reactions (for example, the pentose phosphate shunt).

Aerobic metabolism does not seem to be of great importance in gas gland cells. In histological studies Dorn (1961) and Morris & Albright (1975) found only thin and irregularly distributed mitochondria. Bostrom et al. (1972) measured only a very low activity of cytochrome oxidase in gas gland tissue of the cod, Gadus morhua. This is in line with the low rate of aerobic glucose metabolism in our experiments, which was estimated from the O₂ consumption rate to be 2 nmol min⁻¹ or only 1–2% of the total glucose metabolized.

Swimbladder tissue is known to produce lactic acid even in the presence of oxygen (Ball et al. 1955; D’Aoust, 1970), and its rate of production allows us to estimate the rate of glucose metabolized by this route. Since two lactic acid molecules are formed from each glucose molecule, a Ṁ_La/Ṁ_Glu ratio of 2 would indicate complete degradation of glucose to lactate. The average value in our study of 1–2 indicates that only 60% of the glucose was used for anaerobic glycolysis. Considering also the oxidatively metabolized glucose, about 38% of the glucose metabolized could have entered the pentose phosphate shunt. It is not possible to predict the ratio of CO₂ to glucose in this pathway, but our data seem to indicate that most of the CO₂ was formed by the pentose phosphate shunt.

Anaerobic formation of CO₂ appears to have several advantages for gas secretion into the swimbladder, which requires generation of high gas partial pressures, particularly of O₂ and CO₂, in blood (Fange, 1983). Let us first consider a situation in which high O₂ and CO₂ gas partial pressures are formed in a swimbladder by aerobic CO₂ and anaerobic lactic acid formation, but with no iextra CO₂ production (Fig. 3). would mainly be increased by aerobic CO₂ formation, and countercurrent enhancement in the rete (Kuhn et al. 1963) would further increase ⁠. The ensuing combined respiratory and metabolic acidosis (lactic acid formation) would release O₂ from the haemoglobin bond via the Bohr and Root effects (see Steen, 1970), and the increased would be further enhanced in the rete countercurrent system.

It should, however, be noted that both CO₂ and O₂ secretion into the swimbladder would counteract the increases in their partial pressure. Moreover, any CO₂ molecule formed would consume one O₂ molecule, thus also reducing And, even if some CO₂ were formed from HCO₃⁻, this would reduce the total salt concentration which is required for raising the partial pressures of inert gases (e.g. N₂ and Ar) by the salting-out effect (Gerth & Hemmingsen, 1982; Enns et al. 1967; Pelster et al. 1988a).

In the presence of an anaerobic route for CO₂ formation (Fig. 3), the efficiency of raising all partial pressures would be significantly increased. itself can thereby attain higher values, and this would not reduce levels. The highest values will be found in the gas gland cells, where CO₂ is formed, and CO₂ will diffuse from the tissue to the swimbladder and to the blood (Fig. 3). The concomitantly more pronounced reduction in pH would yield higher ⁠, levels in the blood by the Root and Bohr effect. Furthermore, HCO₃⁻ could be formed from CO₂ despite the lactacidosis, and this would increase the salt concentration for the salting-out effect.

It is thus evident that anoxidative CO₂ formation, for example via the pentose phosphate shunt, may be of great significance for the formation of high gas partial pressures in the swimbladder.

The authors wish to thank Mrs G. Ryfa and Mr S. Rohr for expert technical assistance. Financial support by the Ministerium für Wissenschaft und Forschung des Landes Nordrhein-Westfalen (Grant no. IV B 4-10200687) is gratefully acknowledged.