Many fish species adapt to hypoxia by reducing their metabolic rate and increasing hemoglobin-oxygen (Hb-O2) affinity. Pilot studies with young broods of cichlids showed that the young could survive severe hypoxia in contrast with the adults. It was therefore hypothesized that early exposure results in improved oxygen transport. This hypothesis was tested using split brood experiments. Broods of Astatoreochromis alluaudi, Haplochromis ishmaeli, and a tilapia hybrid (Oreochromis) were raised either under normoxia (NR; 80–90% air saturation) or hypoxia (HR; 10% air saturation). The activity of the mitochondrial citrate synthase was not different between NR and HR tilapia, but was significantly decreased in HR A. alluaudi and H. ishmaeli, indicating lowered maximum aerobic capacities. On the other hand, hemoglobin and hematocrit levels were significantly higher in all HR fish of the three species, reflecting a physiological adaptation to safeguard oxygen transport capacity. In HR tilapia, intraerythrocytic GTP levels were decreased, suggesting an adaptive increase of blood-O2 affinity. Similar changes were not found in HR H. ishmaeli. In this species, however, all HR specimens exhibited a distinctly different iso-Hb pattern compared with their NR siblings, which correlated with a higher intrinsic Hb-O2 affinity in the former. All HR cichlids thus reveal left-shifted Hb-O2 equilibrium curves, mediated by either decreased allosteric interaction or, in H. ishmaeli, by the production of new hemoglobins. It is concluded that the adaptation to lifelong hypoxia is mainly due to improved oxygen transport.
- hypoxia acclimation
- phenotypic plasticity
- oxygen transport
cichlids show extensive species differentiation and, like most fresh water fish, experience large variations in environmental conditions. This makes them excellent models for studying the mechanisms and species specificity of metabolic adaptations to environmental conditions (23, 32, 44). During short-term hypoxia, behavioral and regulatory changes commonly result in decreased energy consumption, increased oxygen extraction, and increased anaerobic metabolism. Fish that are exposed to short-term hypoxia, normally react with increased ventilation, reduction of external activity, and increased aquatic surface respiration (32). When given time to adapt to the new environment, metabolic rates usually decrease resulting in higher hypoxia tolerance (27) and reduced standard metabolic rate (21, 33).
Experiments with immature, as well as with adult fish show that survival through 1 to 2 mo of hypoxia is largely attributable to a reduction in energy expenditure (15, 19, 30, 51, 52). After 6 wk of hypoxia exposure, the oxygen consumption rate of tench (Tinca tinca) and carp, (Cyprinus carpio), decreases to about 50% of that of normoxic individuals (15, 19). Hypoxia-acclimated tench show a 43–76% decrease in the total surface of muscle capillaries and a 60% reduction in mitochondrial volume density, reflecting a reduced aerobic capacity (15). In the hypoxia-tolerant goldfish (Carassius auratus), chronic hypoxia exposure depresses protein synthesis in the liver and elevates the activity of enzymes that promote conservative use of glycogen stores in the muscles (31). In addition, stores of phosphocreatine in the white muscles of this species are significantly increased (30). In young (35 g) carp, chronic hypoxia results in hormonal changes that are associated with retarded gonadal development and reductions in spawning success, sperm motility, fertilization success, hatching rate, and larval survival (51). These observations are compatible with a reduction of metabolic rate, a lower aerobic capacity, and a higher anaerobic capacity.
Most studies on chronic hypoxia pertain to the exposure of (semi)adult fish for periods of 1 wk to 2 mo. Many fish species, however, grow about a 1,000-fold between the postlarval and adult stages. Conceivably, the phenotypic responses might be much larger in early development compared with late juvenile and adult stages. To our knowledge, only a few studies have addressed the effects of lifelong hypoxia exposure (1, 2). Pilot tests with newly hatched young of the cichlid Astatoreochromis alluaudi showed that these fish can adapt to 10% air saturation (AS; ∼2 kPa Po2) levels at 25°C (Van den Thillart GEEJM and Witte F, unpublished data); in fact, in one such experiment, hatchlings of A. alluaudi grew to adulthood at the same rate under hypoxia as under normoxia. In contrast, normoxic adults exposed to 10% AS showed reduced metabolic rates and died within days.
Aquatic vertebrates commonly respond to hypoxia exposure with marked increases in blood hemoglobin concentration (4) and in blood-O2 affinity. Increases in Hb-O2 affinity can be mediated by a fall in organic phosphates of which ATP and GTP are the most important in fish (4, 14, 29, 43, 44, 46). An alternative strategy to modulate oxygen affinity under hypoxia is a change in iso-Hb composition, involving preferential synthesis of high-affinity hemoglobins, or a reduction in the amount of low-affinity hemoglobins. Despite extensive studies, only small alterations in the hemoglobin composition of fish have been observed following short- and long-term hypoxia acclimation (11–13, 20).
Thus, we hypothesized that lifelong exposure to hypoxia, particularly starting with early juveniles, could convincingly demonstrate the existence of major adaptive responses in oxygen transport and energy metabolism. This hypothesis was tested by exposing split broods from three different cichlid species for 15–19 mo to either 10% AS or 80–100% AS.
MATERIALS AND METHODS
Raising and conditioning.
We used broods of three species: tilapia (Oreochromis mossambicus hybridized with O. niloticus), A. alluaudi and Haplochromis (Labrochromis) ishmaeli. The latter two species are haplochromine cichlids. The natural habitats of these species are different. Both parental species of the tilapia hybrid in this study live in habitats with varying oxygen concentrations (48). The tilapia used were F1 offspring of animals obtained from the University of Nijmegen (Nijmegen, The Netherlands).
The habitat of A. alluaudi includes both well-oxygenated streams and hypoxic wetlands (6, 50). Oxygen concentrations in the habitat of H. ishmaeli are at a stable high level and hypoxic events are rare (35, 50). The A. alluaudi and H. ishmaeli specimens studied were offspring of animals caught in the Mwanza Gulf in 1984, which since then have been bred in the laboratory in Leiden. Although A. alluaudi and H. ishmaeli were collected at the same locality, it should be noted that the area was fringed by hypoxic papyrus swamps, which are generally entered by the former, but as far as we know not by the latter. Also we cannot exclude that the responses of both species were affected by the 20 yr in captivity (10–15 generations).
Of A. alluaudi and H. ishmaeli, one brood each was raised during 1999–2000. Of H. ishmaeli and tilapia, one brood each was raised during 2002–2003. To differentiate between the two H. ishmaeli broods, they are named brood 1 and brood 2, respectively. Juveniles were selected when the animals were about 1.5 cm standard length, i.e., about 4 wk after release from buccal cavity. Each nest was split randomly into two groups that were raised under normoxia (NR group) and hypoxia (HR group). The nest size of A. alluaudi was 42, that of H. ishmaeli was 22 and 24 (for broods 1 and 2, respectively). From a large nest of tilapia, 60 just-released juveniles were selected randomly. Each group was split into two parts and raised separately under hypoxia and normoxia. In addition, the tilapia groups were reduced after 6 mo to two groups of 15 by eliminating the smallest and the largest individuals.
All fish were raised for 15 to 19 mo in the same climate room in 45 × 50 × 50-cm glass tanks that had an extra compartment from which water was pumped into the animal compartment to ensure rapid mixing. To follow growth during conditioning, the standard length of all animals was measured at regular time intervals. At adulthood, which occurred at the end of the exposure, survival was almost 100% in both NR and HR groups. The water in the aquaria of the NR groups was kept at 80–90% AS (16–20 kPa Po2). The AS of the water of the HR groups was lowered stepwise to 10% (∼2 kPa Po2) in 4 wk. The fish were kept at 25.5°C and a 12:12-h light-dark cycle. The A. alluaudi and H. ishmaeli were given a diverse diet of flake food, frozen midge larvae, frozen zooplankton, and a mixture of pulverized shrimps, mussels, and flake food. The tilapia were fed with Cyclops and Duplarin (Dupla Aquaristik), and from about 5-cm standard length onwards were fed commercially available 4.5-mm tilapia pellets (Trouw Nutritions).
Hypoxia was obtained by a continuous infusion of degassed water (6–9% AS) at 1–2 l/min. A stainless steel plate that was placed 3 cm below the water level prevented oxygen uptake from the air and surface breathing by the fish. The oxygen level of the water was regulated by Applikon biocontrollers (ADI 1030) equipped with polarographic oxygen sensors (Applikon model ZZ71202AP10) and controlling solenoid valves positioned in sequence with air diffusers. Thus, air bubbling through the water in the extra compartment was automatically activated when the oxygen level fell below the set point, whereby the oxygen level in the animal chambers was kept constant within 1% AS. Water from the same biological filter system was used to continuously refresh the normoxia and hypoxia tanks. The inflowing water of the hypoxia setups was degassed by a vacuum system as described by Van den Thillart and Smit (31).
To minimize stress due to handling, two fish were caught once per day from the same tank, which happened each Tuesday and Thursday between 1000 and 1100. The fish were anaesthetized with MS222 (300 ppm). Animals were completely sedated within 1.5–2 min, and blood was withdrawn within 3 min of sedation. Blood samples were taken from the caudal vein with 1-ml ice-cold syringes containing 10 μl of saline with high heparin content (10,000 IU/ml) and were kept on ice. The saline and blood volumes were determined by weighing. Immediately after blood withdrawal, a sample of white muscle tissue was taken from both sides of the fish between the fifth and tenth dorsal fin ray and above the lateral line. The tissue samples were instantly frozen with freeze clamps cooled in liquid nitrogen. The samples were stored in liquid nitrogen until further processing.
The frozen muscle tissue samples were ground to a fine powder in a mortar that was cooled in liquid nitrogen. The powdered samples were stored in liquid nitrogen for further processing.
Portions of the sample were suspended in nine parts of a 0.1-M KH2PO4 buffer, pH 7.4, vortexed and centrifuged for 10 min at 21,000 g. The supernatant was stored at −80°C and used for measurement of lactate dehydrogenase (LDH), pyruvate kinase, and citrate synthase (CS) activities. The suspensions used for CS activity measurements were sonified for 10 s before vortexing. The LDH activity (EC 188.8.131.52) was determined spectrophotometrically by measuring the conversion of NADH into NAD+ at 340 nm (36). CS (EC 184.108.40.206) was determined spectrophotometrically by measuring the conversion of 3-acetylpyridine adenine dinucleotide (APAD+) into its reduced form (APADH) at 340 nm (26). Glycogen was measured in portions of the ground samples by complete hydrolysis into glucose with gluco-amylase and spectrophotometric determination of the glucose concentration from the conversion of NADP+ into NADPH at 340 nm (17). The concentration of creatine phosphate in the tissue samples was determined by HPLC, following the protocol of Harmsen et al. (8). The concentration of creatine was determined in the same samples according to Wahlefeld and Siedel (39).
Separate aliquots of whole blood were used for the measurement of hematocrit and the concentrations of hemoglobin, ATP, and GTP. Hemoglobin concentrations were measured by a standard spectrophotometric method (Roche). For nucleotide measurements, 50 μl of whole blood was kept on ice for 10 min after adding a mixture of 200 μl perchloric acid (PCA) solution (8% PCA, 10 mM EDTA, 4 mM NaF, 40% ethanol). The samples were sonified, remixed, and resonified for 15 s and centrifuged for 10 min. The supernatant was separated, and 200 μl was treated with 50 μl of a 3 M K3CO2 solution and centrifuged again for 10 min. The supernatant was stored at −80°C for later measurement of GTP and ATP by HPLC following the protocol of Harmsen et al. (8).
Remaining blood was centrifuged for 5 min at 10,000 g at 4°C to separate cells and plasma. The erythrocyte pellet was resuspended in saline and centrifuged for 10 min, and the supernatant was removed. This was repeated twice. The washed erythrocytes and plasma were stored at −80°C for further analysis. Cortisol levels in the stored plasma samples were measured using an enzyme immunoassay (Oxford Biomedical Research).
The washed erythrocyte pellet was thawed and diluted with twice its volume of water. Hemoglobin multiplicity was investigated by thin-layer isoelectric focusing on polyacrylamide gels in the 3–9 pH range, using the PhastSystem of Pharmacia Biotech according to the manufacturer's instructions and, as previously described (25), applying 0.5 μl samples per lane. The separated protein bands were fixed and stained with Coomassie Blue (Amersham BioSciences). Hemoglobins are by far the most abundant proteins in the red cells (>85%), where they occur at concentrations close to their solubility limits, other (household) proteins occur at much lower concentrations. In addition, all but the weakest bands were, before staining, also visible by the naked eye from their reddish color. So, there is little doubt that all bands were, in fact, hemoglobin proteins.
For the oxygen-binding curves, blood samples were pooled per group to obtain enough material. Hemoglobin solutions for measurement of oxygen-binding characteristics were prepared by admixing two volumes thawed red cells with eight volumes of distilled water and one volume of 1 M HEPES buffer, pH 7.5, and centrifugation to remove cell ghosts. The supernatant hemoglobin solution was stripped of organic phosphates and other charged ions using Bio-Rad AG501-X8 mixed-bed resin. The hemoglobin was reduced by dialysis against 10 mM HEPES (pH 7.6 at 5°C) that was equilibrated with nitrogen and carbon monoxide and contained 0.1% sodium dithionite, and four additional dialyses against the same nitrogen- and carbon monoxide-equilibrated buffer without added dithionite. Other preparative procedures were carried out at 0–5°C as earlier described (44). The hemoglobin solutions were frozen at −80°C in small aliquots that were thawed individually for oxygen equilibrium measurements.
Oxygen equilibria of stripped composite hemoglobins from HR and NR group specimens were recorded at 25°C and different pH values in the presence of 100 mM KCl and in the absence and presence of ATP and GTP. For stripping, blood samples had to be pooled per group to obtain enough volume (0.5–1.0 ml). The measurements were carried out using a modified diffusion chamber, where the absorptivity of ultrathin layers of hemoglobin solution is continuously recorded during stepwise increases in the oxygen tension. Equilibrating gas mixtures were prepared from pure (<99.998%) nitrogen, atmospheric air, and oxygen by cascaded Wösthoff gas-mixing pumps (44). The pH values and chloride concentrations in subsamples were measured using a Radiometer BMS Mark 2 glass electrode and a Radiometer CMT10 chloride titrator, respectively. Values of oxygen pressure resulting in 50% saturation of hemoglobin (P50) and cooperativity coefficient at 50% saturation of hemoglobin (n50) were interpolated from Hill plots of data points in the 30 to 70% O2 saturation range.
Data were statistically analyzed with the program SPSS version 12.0 for Windows (SPSS, Chicago, IL). A Student's t-test was used to test for differences in standard length between NR and HR group animals of the same age within each species. Two-way ANOVAs were performed using the factors species and treatment group (normoxia and hypoxia). The original data, as well as the residuals of the two-way ANOVAs, were normally distributed based on Kolmogorov-Smirnov tests with P ≤ 0.05. In case of a species effect, a least significant difference post hoc analysis was performed. When significant interactions between species and treatment group were present, implying that the species responded differentially to the treatment (hypoxia), contrast analyses were performed. The latter analyze 2 × 2 subsets of the total factorial design and show for which species the hypoxia effect was significant.
Conditioning and growth.
Two groups of 21 A. alluaudi (NR and HR) and two groups of 11 H. ishmaeli (NR and HR) were raised during 1999–2000. After 53 and 43 wk, respectively, several deaths occurred due to fighting within a short period of time. In both A. alluaudi groups, the dominant male was removed. At that moment, standard lengths, including dominant males of NR and HR A. alluaudi were 57.8 ± 8.13 and 55.3 ± 4.37 mm, respectively. The standard lengths of NR and HR H. ishmaeli were 55.5 ± 2.4 and 56.4 ± 3.4 mm. Standard lengths, measured during the conditioning, were not significantly different between NR and HR animals (Fig. 1). Similar results were obtained from the growth experiments carried out in 2002 and 2003 with tilapia and H. ishmaeli (not shown). In all experiments, the aggressive behavior of the males started after about 1 yr, when they became mature. As mature males are much heavier (2–4 times) than females and preadults, large variability in weight between individuals occurred in the sampled groups at the end of the experiment (Table 1). However, rather small variation occurred in the period before sexual maturation.
Table 1 presents metabolic data for blood and muscle tissue of the six groups (3 species at 2 conditions). For the glycogen samples, the two-way ANOVA showed a difference between species and an interaction between species and treatment. Post hoc analyses revealed a significant difference between NR A. alluaudi and NR tilapia (means 15.5 vs. 17.6 mM; P = 0.004) and between NR H. ishmaeli and NR tilapia (means 14.2 vs. 17.6; P = 0.023). Contrast analyses indicated that glycogen stores of HR individuals were significantly higher than those of NR siblings in tilapia (means 17.6 vs. 28.3; P = 0.004), but not in A. alluaudi and H. ishmaeli.
There were no significant differences in total creatine stores (the sum of the creatine phosphate and creatine concentration) or LDH activity between the three species, nor between NR and HR siblings of the same species (P > 0.05). CS activity of NR A. alluaudi (mean 2.6 μmoles·min−1·g−1) and NR H. ishmaeli (mean 2.8 μmoles·min−1·g−1) was significantly larger than that of NR tilapia (mean 1.6 μmoles·min−1·g−1), as indicated by the post hoc P values of 0.001 and <0.001, respectively. Additionally, contrast analyses showed that the CS activity of NR animals was significantly higher than that of HR siblings in A. alluaudi (P = 0.050) and H. ishmaeli (P = 0.012).
The cortisol levels of NR A. alluaudi and H. ishmaeli were 46.8 and 45.0 ng/l, respectively; there was no difference between the two species and there were no differences between NR and HR siblings (Table 1). The cortisol levels of NR tilapia were 23.1 ng/l, which was significantly lower than those of A. alluaudi and H. ishmaeli, but also for this species, no differences between HR and HR siblings were found (Table 1). The two-way ANOVA showed that for hemoglobin and hematocrit the values were significantly higher in HR than in NR siblings of all three species (Table 1). No significant differences in the mean cellular hemoglobin concentration were found between NR and HR siblings in any of the three species.
In H. ishmaeli (brood 2) and tilapia, no differences in ATP/Hb were found between NR and HR siblings; nor were there differences between species. In tilapia the GTP/Hb level was significantly lower in HR than in NR siblings (P < 0.001), in contrast to H. ishmaeli, where there was no difference between NR and HR siblings. The same situation was observed for the GTP-to-ATP ratios; in tilapia the ratios were significantly different with 2.17 and 0.94 for NR and HR siblings, respectively, while in H. ishmaeli the ratios were (0.45 vs. 0.36) not significantly different. In addition, GTP/Hb and GTP-to-ATP ratios were much higher in tilapia than in H. ishmaeli.
Thin-layer isoelectric focusing was performed on blood samples of 6–10 individuals of each treatment group. Within each group, no individual variation in iso-Hb patterns was observed. In A. alluaudi and in tilapia, no differences were seen in the number of bands between the NR and HR groups, all fish showing the same iso-Hbs (see Fig. 2A for A. alluaudi and Fig. 2C for tilapia). In H. ishmaeli, however, the NR and HR animals of both nests that were used exhibited strikingly different iso-Hb compositions (hemoglobin switching). Of the total of 14 isoforms that were found in both groups, NR specimens showed nine iso-Hbs, while HR animals showed 10 bands. As shown (Fig. 2B), NR specimens did not express isoforms 2, 7, 9, 11, and 13, whereas HR counterparts lacked isoforms 5, 8, 12, and 14. Of the components that were present in both groups, isoforms 6 and 10 were much more abundant in NR than in HR fish.
The oxygen equilibrium properties, determined to probe the functional consequences of the differences in iso-Hb composition observed in H. ishmaeli (Fig. 2), showed marked differences between the stripped (cofactor-free) hemolysates from HR and NR specimens (Figs. 3 and 4). Under the same experimental conditions, the oxygen affinity was higher in the HR than in the NR hemoglobin [P50 values = 3.48 and 4.50 mmHg, respectively (0.46 vs. 0.59 kPa), at 25°C and pH 7.63]. The difference between HR and NR is strongly magnified by organic phosphates (P50 = 6.7 and 10.2 mmHg) in the presence of saturating levels of ATP, and 8.5 and 12.1 mmHg in the presence of GTP]. In the physiological range (pH 7.6–7.0) the Bohr factor (Δlog P50/ΔpH) was low in the stripped HR and NR proteins (−0.51 and −0.44, respectively), but increased markedly in the presence of either nucleoside triphosphate (to approximately −1.2), indicating a drastic decrease in oxygen affinities, and even greater difference in the affinities between HR and NR group specimens, when the blood becomes acidified in the respiring tissues. Thus, in the presence of phosphates, the P50 difference between HR and NR hemoglobins increases from ∼3.1 mmHg (0.41 kPa) at pH 7.6 to 17 mmHg (2.26 kPa) at pH 7.0 (Fig. 4). The cooperativity n50 was slightly higher in hemolysates of NR fish than in those of HR fish. The HR and NR hemolysates showed the same sensitivity to organic phosphates and pH (as illustrated by the similar phosphate-induced log P50 shifts, and the same slopes of the curves/regression lines, respectively, in Fig. 4).
Both NR and HR fish were active and showed normal social interactions, indicating that the animals were unstressed in both conditions. Mortality due to fighting of dominant males was higher in the NR than in HR groups, which occurred at the end of the conditioning period when the animals became sexually mature. This indicates that hypoxia in the HR groups reduced their activity levels. A similar conclusion followed from feeding behavior; the NR fish fought about the food in contrast to their HR siblings. After about 1 yr, when the animals sexually matured, large differences in mass occurred between dominant males and the other individuals from the same group, which led to differences in mass between the sampled groups in the end, which was significant for A. alluaudi only (Table 1). However, growth rates (both length and weight) up to adulthood were in all cases the same for NR and HR siblings (Fig. 1).
The cortisol level of unstressed fish is generally below 100 ng/l (37); the low levels observed in our study (<47 ng/l) therefore suggest low-stress conditions. It has been observed, that exposure of tilapia to stepwise decreases in AS levels below 20%, raises cortisol concentrations (38). Cortisol has a suppressing effect on the immune system and is associated with retarded growth (49), so high chronic cortisol levels would have been detrimental. However, neither growth rates nor cortisol levels were different between HR and NR siblings. We therefore contend that 10% AS was not a stressful condition for hypoxia-raised cichlids in this study.
In general, glycogen stores in liver and white muscle appear to be higher in fish species that are anoxia/hypoxia-tolerant, such as tench (T. tinca), goldfish (C. auratus), and Crucian carp (C. carassius). In contrast, intolerant species, such as rainbow trout (Oncorhynchus mykiss) and cod (Gadus morhua), have very low levels of muscle glycogen (34). In A. alluaudi, H. ishmaeli, and tilapia, the white muscle stores of glycogen were ∼14–28 μM glucose/g tissue, which is in the same range as in the above-mentioned hypoxia/anoxia tolerant species. Compared with their NR counterparts, glycogen concentrations in the white muscles of HR specimens were significantly elevated in tilapia, but not in HR A. alluaudi and H. ishmaeli (Table 1). This suggests that HR tilapia compensate limitations of the oxygen extraction by increasing the anaerobic capacity of their muscles. An elevated anaerobic metabolism may be expected to be associated with an elevated LDH activity, which was not observed. Greaney et al. (5) argued that white muscles have a high glycolytic capacity and therefore may be preadapted to hypoxia and lack the need for a capacity increase during chronic hypoxia. On the other hand, an increase of LDH is only required for a higher flux (during peak activities), whereas an increase in glycogen with the same LDH activity suggests longer anaerobic endurance, which may be more advantageous under chronic hypoxia.
In HR A. alluaudi and H. ishmaeli, CS activity was significantly reduced compared with NR individuals. However, the decreased CS activity did not affect routine activity levels, since oxygen consumption levels of HR cichlids were the same or higher than those of NR siblings (Table 2). CS was measured at standard conditions, which do not reflect in vivo situations. Also the tissue conditions of hypoxia- and normoxia-acclimated fish may have been different. Still, CS is, as part of the TCA cycle, located within the mitochondria and is often used as a mitochondrial marker enzyme. In humans, as well as in other mammals, high altitude acclimation corresponding with low oxygen levels causes a reduction in muscle mitochondrial content together with a reduction of all mitochondrial enzymes (10). Similarly, Johnston and Bernard (15) showed that tench acclimated to hypoxia (8.5% AS) for 6 wk have a decreased mitochondrial density in the white muscles. Thus, it is plausible that the decreased CS activity in A. alluaudi and H. ishmaeli was due to a reduction in mitochondrial density of the white muscles. Maximal aerobic capacity in fish is, however, about 10 times higher than the standard metabolic rate. So, a reduction in CS will cause a reduction in maximal aerobic activity, but is unlikely to affect the routine oxygen consumption.
Vertebrates commonly respond to hypoxia with increases in the concentration and oxygen affinity of the hemoglobin. In vertebrates, organic phosphates are important allosteric factors and provide a rapid means of adapting hemoglobin function to ambient oxygen tension and tissue oxygen demand (15, 22, 28, 29, 43, 46). Fish use polyanionic nucleoside triphosphates that bind in a positively charged cavity between the β-chains of the hemoglobin. While all fish have high erythrocytic ATP levels, some species also have a significant amount of GTP, which binds in the phosphate pocket with an additional hydrogen bond compared with ATP and decreases Hb-O2 affinity more potently (7, 43, 45). Although the distribution of GTP in teleosts defies strict characterization in terms of phylogeny and oxygen availability (18, 29, 44), general trends are discernable in teleosts. Thus, GTP-to-ATP ratios are low in salmonids that frequent oxygen-rich waters and in benthic pleuronectiforms (flatfish), which are inactive but high in anguilliforms (eels) (44) and cichlids (this study) that tolerate hypoxic conditions.
The red blood cells of NR tilapia used in this study contained about twice as much GTP as ATP (Table 1). The HR tilapia showed a 55% reduction in GTP levels, resulting in equimolar ATP and GTP concentrations. In contrast to tilapia, the HR H. ishmaeli showed no significant reduction in GTP levels (Table 1). This supports the view that increase in Hb-O2 affinity is regulated differently in this species.
Hemoglobin multiplicity is a normal phenomenon in fish; its function is, however, little understood. In a large survey on hemoglobins of 77 Amazon fish, Fyhn et al. (3) showed an average of four different hemoglobins per species. Of the 16 cichlid species investigated, the numbers ranged between five and nine different hemoglobins. A few species showed polymorphism; however, in most species, the patterns were species specific (3). Also our data showed no polymorphism, the isoelectric focusing gels of the blood samples from the same group showed exactly the same pattern. For A. alluaudi and tilapia we found 14 and 9 different bands, respectively (Fig. 2). Isoelectric focusing of the hemolysates of A. alluaudi and tilapia showed no differences in hemoglobin multiplicity between NR and HR siblings (Fig. 2, A and C, respectively). However, both broods of H. ishmaeli exhibited clear differences in iso-Hb composition between NR and HR siblings (Fig. 2B). HR fish lacked four iso-Hbs that were present in NR siblings. Analogously, five new hemoglobins were seen in HR specimens that were lacking in NR H. ishmaeli. This pattern was consistent for all NR (n = 13) and HR (n = 17) fish examined. Such a clear-cut difference has, to our knowledge, not earlier been observed in fish. The distinct hemoglobin pattern in hypoxia-raised H. ishmaeli, suggests that it constitutes part of a regulatory mechanism induced by ambient oxygen availability. This tallies with the oxygen-binding data that show consistently higher Hb-O2 affinities in HR than in NR hemoglobins under a range of different measuring conditions (four pH values and in the absence and presence of ATP and GTP; Figs. 3 and 4). All else equal, this translates to a higher Hb-O2 affinity in the animals that were subjected to lifelong hypoxia.
The similar sensitivities of the HR and NR hemolysates to organic phosphates and pH indicate that the amino acid substitutions that underlie the differences in electrophoretic mobility do not involve amino acid residues implicated in oxygenation-linked proton or phosphate binding (unlike the electrophoretic-cathodal salmonid hemoglobins that have lost Bohr and ATP sensitivities). Instead, the substitutions alter the intrinsic oxygen affinity, and they thus are likely to alter the oxygen-linked hydrogen bonds or salt bridges, that (de)stabilize the tense (low-affinity, deoxy-Hb) or the relaxed (high-affinity, oxy-Hb) structure of the molecules. This contention is supported by the consistent (albeit small) difference in the cooperativity coefficients between the NR and HR hemoglobins (Fig. 4). Remarkably, the higher intrinsic oxygen affinity observed in the HR H. ishmaeli is not matched by lower ATP-to-Hb and GTP-to-Hb ratios than in NR specimens. This suggests that the new hemoglobins in the HR group introduce a new oxygen affinity set point that can be increased even further under specific in vivo conditions, e.g., following hyperventilation and adrenergic stimulation of the red cells (22).
Table 2 gives a summary of the different phenotypic changes observed in the three cichlid species. Very remarkable is the fact that NR and HR siblings of all three species grew at the same rate (during their juvenile stage). With respect to metabolism, there was an increase in muscle glycogen in tilapia but a decrease in citrate synthase in A. alluaudi and H. ishmaeli. The first effect suggests improved anaerobic capacity, while the second shows a decreased aerobic peak capacity. These effects correspond with the observation that the HR siblings were slower in their movements. Furthermore, we observed three phenotypic responses that enhance oxygen uptake under chronic hypoxia (Table 2). First, a higher erythrocyte level under hypoxia was demonstrated in all three species. This is likely the result of an hypoxia-stimulated HIF-1 production that leads to the expression of many proteins including erythropoietin. Increased erythropoietin levels stimulate the erythrocyte production. In that respect, hypoxia effects are similar to high altitude effects. Second, a decrease in the concentration of GTP in the erythrocytes of tilapia was found. Increasing the Hb-O2 affinity by altered concentrations of modulators, such as ATP and GTP in lower vertebrates and 2,3-diphosphoglycerate in mammals is a commonly found strategy. The possibility of a left shift of the Hb-O2 dissociation curve is crucial for all animals when they encounter Po2 levels that are too low to saturate the hemoglobin. Such a physiological response is very relevant in environments with variable oxygen concentrations. Third, HR H. ishmaeli showed different iso-Hbs with increased oxygen affinity. In contrast to altering GTP concentrations, this response is effective under long-term or permanent hypoxia. Changes in hemoglobin multiplicity within the same individuals as a response to changed environmental conditions, are almost solely related to transitions in early life stages, for instance during the water-to-air developmental transition in air breathing fish or amphibians or after birth in mammals or viviparous fish (41, 42). Therefore, a possible hypothesis is that the presence of different hemoglobins in NR and HR H. ishmaeli is an example of heterochrony. Appearance of new hemoglobins under chronic hypoxia in H. ishmaeli, resulting in increased blood-O2 affinity clearly suggests a new and unique way of adaptation to hypoxic environments.
The habitat of H. ishmaeli before the recent spells of lake-wide chronic hypoxia (9, 40) consisted of shallow (<6 m), open waters with stable normoxic conditions (35, 50) that provided no apparent selective force to become hypoxia tolerant. However, the desiccation of Lake Victoria ∼15,000 years ago (16) may have favored the development of hypoxia tolerance in its evolutionary past, when the area consisted of swamps.
We thank Anny Bang for skilled technical assistance and the Danish Natural Science Research Council for support. We also thank Patrick Niemantsverdriet for animal care, Harald van Mil for help with the statistical analyses, and the three anonymous referees for their constructive comments.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2007 the American Physiological Society