The Generation of Hyperbaric Oxygen Tensions in Fish

Bernd Pelster


Surprising inventiveness in the molecular interactions in fish hemoglobins that express the Root effect (decreased oxygen-carrying capacity at low pH) and in metabolic adaptations in swim bladder gas gland cells and retinal tissues causes local acidification of blood and generates hyperbaric oxygen tensions that drive oxygen into the swim bladder (regulating buoyancy) and ensures the oxygen supply to the avascularized retinae.

Aerobic metabolism requires a supply of oxygen to the tissues and adequate elimination of acid end products like carbon dioxide to prevent undue acidification of tissues. Accordingly, metabolic and ventilatory activities are tightly coupled in most animals. In addition, the low solubility of oxygen in body fluids is compensated by the presence of respiratory pigments like hemoglobin or hemocyanin, which increase the oxygen-carrying capacity of blood or hemolymph by one to two orders of magnitude. The optimal use of these respiratory pigments often requires loading and unloading of the oxygen in a very narrow range of oxygen partial pressures. Furthermore, variations in lifestyle, environmental conditions, and internal organization require accurate fine-tuning of the oxygen-binding characteristics of the respiratory pigments.

Intensive studies on the structure and functions of the best-known respiratory pigment, hemoglobin, have revealed a striking flexibility in its intrinsic oxygen-binding characteristics even among closely related species, caused by replacements in various amino acid residues of the protein moiety of these pigments (4). In addition, the hemoglobins may show large variability in their sensitivity to effectors like protons, organic phosphates, or bicarbonate, which commonly bind to the deoxygenated state of the pigment and reduce its oxygen affinity.

An increase in proton concentration decreases the oxygen affinity of most respiratory pigments (the well-known Bohr effect). In a large number of fish hemoglobins, however, pH decrease not only reduces oxygen affinity but also reduces oxygen-carrying capacity (Fig. 1) (2, 16). At low pH, these hemoglobins remain partially deoxygenated even in the presence of oxygen partial pressures in excess of 20 kPa (i.e., under severe hyperoxia). Although some scientists view this as an extreme form of the Bohr effect, others consider it to be a specific phenomenon with a distinct molecular mechanism. It was named the “Root effect” after R. W. Root, who first described this pH-dependent decrease in hemoglobin oxygen-carrying capacity.


An increase in proton concentration decreases hemoglobin oxygen affinity in most vertebrate hemoglobins (the Bohr effect). In fish blood, an increase in proton concentration may also induce a decrease in the hemoglobin oxygen-carrying capacity (the Root effect). Because of specific amino acid replacements in the α- and β-chains, the oxygenated conformation (R-state) of the Root effect hemoglobins is destabilized at low pH, whereas the α1β2 switch interface in the deoxygenated T-state of the molecule is stabilized, whereby the hemoglobin molecules cannot be oxygenated at low pH even at high oxygen partial pressures (Po2). Co2, oxygen content.

Although the exact molecular mechanism basic to the Root effect is still debated, it is widely accepted among fish physiologists that the Root effect is required to generate hyperoxic oxygen partial pressures in the swim bladder and eyes of fish. These high oxygen partial pressures are necessary to fill the swim bladder, especially of physoclist fish (whose swim bladders are not connected to the esophagus after the embryonic stage), with gas and to ensure a sufficient oxygen supply to the retina.

Presence and nature of the Root effect

The presence of a Root effect in blood can be investigated gasometrically and spectrophotometrically by measuring the oxygen content or the oxygen saturation, respectively, of blood or hemoglobin solutions at various pH values. These different approaches may imply variations in the oxygen partial pressure used as a reference for 100% oxygenation, which have contributed to controversies about the distribution of the Root effect. Nevertheless, it is generally accepted that the Root effect is a peculiarity of teleost fish hemoglobins, although a few studies have reported it in elasmobranchs and even in amphibians (13). The possible physiological function of the Root effect in nonteleosts remains obscure and is not dealt with in this review, which focuses on teleost fishes.

Fish hemoglobins can be separated into electrophoretically anodic and cathodic components, and the relative distribution of these hemoglobin components varies from species to species (17). Although the Root effect is clearly established in the anodic components, it is absent in cathodic hemoglobins (13). Accordingly, the magnitude of the Root effect varies from species to species. In eel or trout, for example, in which ~50% and ~60%, respectively, of the hemoglobin cannot be oxygenated at low pH, these values match with the fraction of anodic hemoglobins.

The molecular basis of the Bohr effect is a binding of protons to specific amino acid residues of the globin chains that results in extreme stabilization of the low-affinity, deoxygenated conformation (T-state) of the hemoglobin. Consequently, higher oxygen tensions are required to oxygenate the hemoglobin, i.e., its oxygen affinity is decreased. On the basis of the concept that the Root effect is distinct from the Bohr effect, it was proposed that specific amino acid replacements (14) in Root hemoglobins are implicated in blocking full oxygenation at low pH even under high oxygen partial pressures.

Root hemoglobins appear to exhibit marked chain heterogeneity and characteristically show negative cooperativity (the Hill coefficient is below unity at low pH). Recent molecular studies suggest that protonation of specific positively charged residues within the central cleft between the two β-chains causes a destabilization of the oxygen conformation (R-state) at low pH (6). Additional amino acid replacements (Table 1) appear to make the α1β2 switch interface, which is loosened during oxygenation, particularly stable at low pH, so that Root hemoglobins can bind oxygen at the high-affinity α-chains while remaining in the T-state conformation. This ability to accommodate conformational changes in liganded subunits while remaining in the T-state is not observed in hemoglobins showing a normal Bohr effect but no Root effect (3). The molecular mechanisms hitherto proposed based on single amino acid exchanges have not been supported by experimental data, and Mazzarella et al. (5) therefore proposed that more than one amino acid constellation might cause the presence of the Root effect. This might imply that the Root effect was invented more than once during evolution.

View this table:

Amino acids that are thought to be involved in the expression of the Root effect

Blood is acidified by secretion of protons into the blood or through increases in carbon dioxide partial pressure. Reoxygenation of hemoglobin is achieved by alkalinization of the blood. The kinetics of the Root-on and Root-off reactions are rather complex because they not only include the chemical reaction and the diffusion of oxygen but also the transfer of acid into the blood and erythrocytes and binding of protons to the hemoglobin (12). Carbonic anhydrase is found in the red blood cells and is involved in both processes. The pH dependence of their hydration and dehydration reactions significantly contributes to differences in the kinetics of the Root-on and Root-off reactions. Nevertheless, both reactions are completed within <1s, and these differences in the kinetics do not appear to be of physiological significance.

Physiological importance of the Root effect

Given that the Root effect is switched on by acidification, we have to address the question of when and where acidification occurs. Acidification of blood is observed in specific stress situations, such as strenuous exercise, during which the anaerobic production of lactic acid in muscle tissue results in a severe drop in the pH of venous and even arterial blood. Acidification of arterial blood, however, reduces the oxygen-carrying capacity of the blood in the gills, where it cannot be fully oxygenated. This would be deleterious in situations in which as much oxygen as possible is required for muscular contraction. Intriguing ion transport mechanisms in the membranes of the erythrocytes ensure that this lactic acidosis is not transferred to the cytoplasm (11).

A sodium/hydrogen exchanger (NHE) is present in most erythrocytes and is usually involved in osmotic volume regulation. In fish with Root effect hemoglobin, this exchanger is under β-adrenergic control (β-NHE; Ref. 8). Stimulation of the β-adrenergic receptor via cAMP as a second messenger and protein kinase A activates the NHE, which extrudes protons from the red blood cells and thus alkalinizes the cells and acidifies the blood plasma. Thus intracellular pH of the red blood cells may even increase under exercise-induced blood acidification.

These considerations show that, physiologically, the Root effect is not tailored to bring about an overall reduction in oxygen-carrying capacity of the blood. To understand the physiological importance of the Root effect, we therefore turn our attention to special organs. Only the swim bladder and the eye of the fish have been identified as organs that rely on the Root effect. Both organs depend on a buildup of oxygen partial pressures that far exceed the 20 kPa that, under optimal conditions, can be reached in the gills. In combination with a sophisticated countercurrent arrangement of the vascular system, oxygen unloading via the Root effect allows for the generation of hyperbaric oxygen partial pressures of several hundred atmospheres in the swim bladder.

Swim bladder function

The swim bladder originates ontogenetically as an outgrowth of the esophagus. Although it is not clear whether it originally was a buoyancy organ or a structure involved in aerial gas exchange, its homology to the lung is generally accepted. The teleost swim bladder primarily functions as a buoyancy organ, and, being flexible, its volume and internal pressure change with changes in hydrostatic pressure. To retain neutral buoyancy, teleosts must be able to keep the volume of the swim bladder constant. This means that gas needs to be secreted into the bladder while descending and removed while ascending. Removal of gas occurs by releasing gas bubbles via the esophagus or by reabsorbing gas along gas partial pressure gradients in special sections of the swim bladder. The secretion of gas into the swim bladder is a more complicated phenomenon that includes the release of oxygen from the hemoglobin via the Root effect. The swim bladder gas consists mainly of oxygen, carbon dioxide, and inert gases, mainly nitrogen. This present discussion is limited to the secretion of oxygen, which is the major constituent. Other gases have been dealt with elsewhere (10, 11).

Although the commonly used term “gas secretion” implies an active process, swim bladder gas filling occurs by simple diffusion along partial pressure gradients initiated by the release of oxygen from the hemoglobin via the Root effect, typically when extracellular pH values fall to ~7.3–7.5 (and intracellular values are ~7.1–7.3). The acidification of the blood within the swim bladder capillaries is brought about by the release of acid from the so-called gas gland cells of the swim bladder epithelium (Fig. 2). These cells are cuboidal or cylindrical and may be lumped together, forming a compact, sometimes multilayered gas gland, or may be spread over the whole swim bladder epithelium, as in the eel. Gas gland cells show structural and functional polarity. The apical side, oriented toward the swim bladder lumen, is characterized by a few small microvilli. The basal side, facing the blood vessels, shows extensive membrane foldings. In contrast to most other cells that carry out active transport, gas gland cells apparently are not equipped with a large complement of mitochondria.


Lactic acid and carbon dioxide produced by gas gland cells are released into the blood stream. The increased extracellular proton concentration dehydrates bicarbonate, forming carbon dioxide. In gas gland tissue, this reaction is catalyzed by membrane-bound carbonic anhydrase. Carbon dioxide diffuses into the red blood cell, where the increase in carbon dioxide partial pressure results in hydration of carbon dioxide, forming bicarbonate and protons. The increase in intracellular proton concentration initiates the Root effect, and oxygen is released from the hemoglobin. CA, carbonic anhydrase; G-6-P, glucose-6-phosphate; Hb, hemoglobin; PPS, pentose phosphate shunt.

Recent studies have revealed that this structural polarity is functionally important. Gas gland cells serve a dual function, producing surfactant, which is secreted into the swim bladder lumen via exocytosis (15), and secreting acidic metabolites into the blood at the basolateral membranes. Acidic metabolites are produced mainly from blood-borne glucose. A large fraction of this glucose is converted into lactic acid, even though gas gland cells are usually exposed to hyperoxic conditions. This is possible because these cells do not show a Pasteur effect and thus produce lactate even at an oxygen pressure of 50 atmospheres. The lactic acid is released into the blood. Although pharmacological inhibition of lactate transport (using cinnamate) does not impair the acid release, inhibition of sodium/hydrogen exchange, anion exchange, or of vacuolar (V)-ATPase activity significantly reduces acid release from the cells (9). Sequence data on the B subunit of V-ATPase of eel gas gland cells has revealed the presence of two isoforms (7). In mammals, one of these isoforms is found in very few organs, such as kidney. In kidney cells, its function in association with acid-base regulation is partially controlled by being inserted or removed from the membrane. The second V-ATPase B subunit isoform (brain isoform) can be found in intracellular vesicles of virtually all cells. If this would be a general model that could be transferred to the gas gland cells, the kidney isoform of the B subunit might be involved in acid secretion at basolateral membranes and the brain isoform may be necessary for the acidification of lamellar bodies, in which surfactant is stored before exocytosis.

Thus several mechanisms appear to cause the release of protons in the gas gland. The pH of blood circulating through the swim bladder of the European eel has been found to be in the range of 7.5–7.8 when the gas gland cells are inactive. The blood becomes acidic, with pH values of 6.5–6.6 at times when the gas gland is active and gas is being secreted into the swim bladder. The presence of different mechanisms for proton transfer across the cell membrane may be related to the fact that the cells must provide for a well-controlled secretion of protons over this wide range of pH values.

The transfer of protons through the cell membrane acidifies the extracellular fluid. This acid load is transferred into the erythrocyte via the Jacobs-Stewart cycle. Extracellular protons react with bicarbonate, forming carbon dioxide that diffuses into the red blood cell, where it is hydrated to bicarbonate and protons. With an increasing proton concentration within the red blood cell, the threshold for the Root effect is reached and oxygen is liberated from the hemoglobin. In swim bladder tissue, the time necessary to complete the whole series of reactions is significantly reduced due to the presence of a membrane-bound carbonic anhydrase in gas gland cells. Although blood plasma typically is devoid of carbonic anhydrase activity, membrane-bound enzyme rapidly accelerates the equilibrium of the carbon dioxide-bicarbonate reaction in the extracellular fluid (9). The high carbonic anhydrase activity of the red blood cell reestablishes the intracellular equilibrium of the bicarbonate-carbon dioxide reaction.

Besides lactic acid, carbon dioxide is produced and released from gas gland cells and is especially important for the initiation of the Root effect. A comparison of the oxygen consumption and the carbon dioxide production of eel gas gland cells revealed that most of the carbon dioxide is produced by decarboxylation in the pentose phosphate shunt and that only a small fraction of the glucose is actually oxidized by aerobic metabolism (10). This intriguing metabolic design ensures that the highest carbon dioxide partial pressure is always found in the gas gland. In contrast to the protons, which cannot penetrate membranes easily and require special, time-consuming transport mechanisms, carbon dioxide rapidly diffuses from the gas gland cells into the red blood cells. Thus the carbon dioxide produced ensures a very rapid acidification of the erythrocytes and initiation of the Root effect. The oxygen that has been liberated from the hemoglobin as blood passes through the gas gland readily diffuses into the swim bladder lumen along the partial pressure gradient.

Some of the oxygen will also remain in the blood, and blood returning from the gas gland tissue travels through the countercurrent system of the rete mirabile with a higher oxygen partial pressure than afferent blood traveling to the gas gland tissue. Given the production and release of carbon dioxide from the gas gland cells, this also applies to carbon dioxide tensions. The metabolic activity of gas gland cells thus brings about initial increases in blood oxygen partial pressures and carbon dioxide partial pressures.

The swim bladder rete mirabile represents a very efficient countercurrent exchanger. Oxygen and carbon dioxide molecules returning to the venous side of the rete mirabile will therefore diffuse back to the arterial side, resulting in a countercurrent multiplication of the initial increases in oxygen and carbon dioxide partial pressures. The back diffusion of carbon dioxide again is of crucial importance for the initiation of the Root effect within the swim bladder tissue because it results in an acidification of the arterial blood of the rete mirabile, induces the Root effect, and elevates oxygen partial pressure in the blood supplying the swim bladder tissue. The metabolic activity of gas gland cells causes further acidification, which enhances the Root effect and the increase in oxygen partial pressure. Model calculations demonstrate that this dual acidification, together with the very efficient countercurrent multiplication of the increases in blood oxygen partial pressure and blood oxygen content in the rete mirabile, can readily explain the generation of oxygen partial pressure values of up to several hundred atmospheres in swim bladder tissue of deep sea fish (10).

The fish eye

In contrast to most other vertebrates, the retinas of most fish are not vascularized, and diffusion distances for nutrients and oxygen to the retina therefore can be more than six times larger than in primates. Because the retina has a high metabolic activity, this raises a question about how the necessary oxygen supply to the retina is secured. Although the situation is far from understood, obvious parallels exist to the situation in the swim bladder, and the Root effect appears to be involved. The vitreous humor of the fish eye is highly oxygenated, and oxygen partial pressure values of 1 atmosphere and higher have been reported (11, 18). Furthermore, the ophthalmic artery passes through a countercurrent system, the choroid rete mirabile, before branching to the choriocapillaris, the capillary network from which the retina is supplied.

To initiate the Root effect, blood must be acidified in the choriocapillaris. As with the gas gland tissue, teleost retinal tissue produces lactic acid even in the presence of oxygen. Furthermore, in addition to aerobic glucose oxidation, part of the glucose is metabolized in the pentose phosphate shunt. Thus, as in the swim bladder, the retina apparently produces lactic acid and carbon dioxide, which are released into the blood stream and initiate the Root effect, and the initial increase in oxygen partial pressure is subsequently multiplied by back diffusion of oxygen in the countercurrent system of the choroid rete mirabile (Fig. 3).


Blood supply to the fish eye originates from the efferent artery of the pseudobranch, which in turn receives arterialized blood from the first gill arch. According to current understanding, blood is titrated in the pseudobranch to a pH value just above that which induces the Root effect. Subsequently, only a small amount of acid needs to be released from the retina to switch on the Root effect in the choriocapillaris and to liberate oxygen from the hemoglobin. The resulting increase in oxygen partial pressure induces a back diffusion of oxygen in the choroid rete mirabile and thus generates high oxygen partial pressures that are necessary to ensure the oxygen supply to the avascularized fish retina. O2cap, oxygen carrying capacity. Figure modified from Ref. 1.

Despite the homologies with the swim bladder, an intriguing twist complicates the situation in the fish eye, since retinal tissue is very sensitive to acidification and acidification down to pH 6.4 results in blindness. Thus the unavoidable drop in blood pH associated with the countercurrent concentration of carbon dioxide must be moderate to avoid an acid-induced damage of the tissue.

Comparative studies demonstrated that the presence of a choroid rete mirabile in the eye coincides with the presence of a pseudobranch (1). This gill-like hemibranch receives arterialized blood from the first gill arch, and the efferent artery gives rise to the ophthalmic artery. The fish eye thus receives its blood supply from the pseudobranch. Although the pseudobranch has attracted attention for more than 150 years, its physiological function remains a mystery. Nevertheless, most fish physiologists subscribe to the idea that it is implicated in the oxygen supply to the retina.

It has recently been proposed that the pseudobranch is equipped with a sensor and titrates the blood to a pH value just above the value at which the Root effect is initiated (1). In this way, only a small amount of acid needs to be released to switch on the Root effect during passage of the choriocapillaris, and the countercurrent concentration of acid, especially of carbon dioxide, is reduced. This scenario indeed could build up high oxygen partial pressure values with only a minor acidification of the blood. In this context, it is important to keep in mind that, in contrast to the swim bladder, where the oxygen partial pressure necessary to drive diffusional oxygen transport into the swim bladder lumen increases with increasing water depth, the choriocapillaris oxygen partial pressure must only be high enough to meet the oxygen requirement of the retina. At this point, no further increase in oxygen partial pressure is necessary. Accordingly, the length of the capillaries of the choroid rete, which is crucial for the maximal countercurrent multiplication that can be achieved, is short compared with the length of capillaries in most swim bladder retia mirabilia.


The Root effect represents an intriguing example of the ecophysiological variability and adaptability of hemoglobin. Although hyperoxia is deleterious for most tissues, extremely high oxygen partial pressures must be generated in the swim bladder and retinae of fish to ensure their proper physiological function. These high oxygen partial pressure values are generated in a complex system that includes a remarkable capacity of certain cells to produce and secrete acid metabolites, the presence of Root effect hemoglobins, and of a rete mirabile. The rete mirabile is necessary for countercurrent multiplication but also acts as a barrier to prevent the acid from spreading through the whole organism. In addition, hormone-induced acid-base regulation in red blood cells prevents the initiation of the Root effect during situations of general acidosis, thus preventing a reduction in hemoglobin oxygen-carrying capacity that would curtail oxygen transport.


I would like to thank Dr. R. Weber for helpful comments on the manuscript.

Parts of this study were financially supported by the Austrian Science Foundation (FWF, P-11837-BIO, and P14174-BIO).


View Abstract