Effects of long-term hypoxia on enzymes of carbohydrate metabolism in the Gulf killifish, Fundulus grandis

Because of the rich evolutionary history and current ecological diversity of teleost fish, the study of this group has been particularly informative in elucidating the responses of animals to hypoxia(Nikinmaa and Rees, 2005). These responses include behaviors to avoid hypoxic areas, utilize well-aerated microenvironments, or reduce activity(Kramer, 1987; Van den Thillart et al., 1994; Dalla Via et al., 1998; Wannamaker and Rice, 2000);physiological and morphological adjustments that improve the oxygen extraction and delivery to tissues (Jensen et al.,1993; Sollid et al.,2003); and biochemical changes that increase the capacity of tissues to function and survive at low oxygen(Hochachka, 1980; Van den Thillart and Van Waarde,1985; Hochachka and Somero,2002). In general, this suite of responses serves to compensate for the decrease in environmental oxygen availability, and results in hypoxia tolerance that, in some species, is quite marked.

One biochemical response to hypoxia is an increase in anaerobic ATP production, typically via glycolysis(Van den Thillart and Van Waarde,1985; Dalla Via et al.,1994; Virani and Rees,2000). A number of studies indicate that hypoxic exposure increases the activities of glycolytic enzymes that presumably augment the capacity of fish tissues for anaerobic energy production(Greaney et al., 1980; Johnston and Bernard, 1982; Van den Thillart and Smit,1984; Dickson and Graham,1986; Lushchak et al.,1998; Zhou et al.,2000; Kraemer and Schulte,2004). However, these increases were not uniformly observed among glycolytic enzymes, across tissues, or among species. For example, hypoxic exposure of killifish, tench, and goldfish led to increased activities of some glycolytic enzymes in liver, but not in white skeletal muscle(Greaney et al., 1980; Johnston and Bernard, 1982; Van den Thillart and Smit,1984). The opposite trend, increased enzyme activities in muscle and no changes in liver, was observed in Hoplias microlepis(Dickson and Graham, 1986). In tissues of other species, enzyme activities stayed the same(Shaklee et al., 1977; Driedzic et al., 1985) or decreased during hypoxic exposure(Almeida-Val et al., 1995). In addition, all but one of the above studies report data on only a subset of the enzymes of glycolysis (as few as one or two). The single study that measured all glycolytic enzymes did so in only one tissue, liver, and found increased activities of enzymes catalyzing reactions close to equilibrium but not for those catalyzing reactions far from equilibrium(Kraemer and Schulte,2004).

The goal of the current study was to generate a comprehensive, multi-tissue perspective of the effects of chronic hypoxic exposure on carbohydrate metabolism in the Gulf killifish Fundulus grandis. F. grandis is a common inhabitant of estuaries along the Gulf of Mexico, areas which may become hypoxic on a daily or seasonal basis. In acute laboratory exposures,aerobic metabolism of this fish decreases below a critical oxygen tension of approximately 4.5 kPa (Virani and Rees,2000). Below the critical oxygen tension, these fish rely upon anaerobic glycolysis to compensate, at least in part, for the reduced energy provision by aerobic metabolism (Virani and Rees, 2000). In the closely related F. heteroclitus,hypoxic exposure of several days to several weeks leads to increased activities of selected glycolytic enzymes in liver(Greaney et al., 1980; Kraemer and Schulte, 2004). We have measured the maximal activities of all glycolytic enzymes in four tissues(white skeletal muscle, liver, heart and brain) of F. grandis after being held under hypoxic or normoxic conditions for 4 weeks. The hypoxic concentration of dissolved oxygen used (1.3 mg l^-1, ∼3.6 kPa)is below the critical oxygen tension for this species, but is ecologically relevant and well tolerated by this species. To get a more complete picture of carbohydrate metabolism, we also measured the activities of enzymes of glycogen metabolism (in muscle and liver) and gluconeogenesis (in liver).

These measurements allowed us to address the following questions about the effects of chronic hypoxia on the metabolic potential of various tissues in F. grandis. Do different tissues respond similarly to hypoxic exposure? Do all enzymes within a given pathway (glycolysis) change in concert(i.e. in the same direction and by the same magnitude)? Finally, in a single tissue (liver) do the capacities for catabolic and anabolic processes respond similarly or differently to chronic hypoxia? The results demonstrate that when intact fish are acclimated to prolonged hypoxia, enzyme activities respond in a tissue-specific fashion and that within a tissue changes in enzyme activities are not uniformly distributed across a metabolic pathway.

Fundulus grandis Baird and Girard 1853 were purchased in late May 2003 at a commercial bait shop in Pascagoula, MS, USA and transported to the Gulf Coast Research Laboratory in Ocean Springs, MS. All fish were habituated to laboratory conditions for 2 weeks, after which fish were randomly allocated to normoxic (N=24) or hypoxic (N=24) treatments. Exposures were carried out in four tanks, each measuring 48 cm×48 cm(length×width) with an overflow drain at 17.5 cm depth (∼40 l). Two normoxic tanks received water from an aerated reservoir and two hypoxic tanks received water from a mixing box with input from aerated and nitrogen-sparged reservoirs. Each tank was divided into four quadrants (each 24 cm×24 cm)by mesh, and each quadrant held one male and two female F. grandis. The 40 l tanks were covered and the partial pressure of oxygen in the air-space above the water level (∼2.5 cm) was essentially in equilibrium with the water. Treatments lasted 4 weeks, during which time dissolved oxygen(DO) was measured twice daily with a Yellow Spring Instruments (Yellow Springs, OH, USA) oxygen electrode. DO averaged 6.68±2.1 mg l^-1 in the normoxic tanks (mean ± 1 s.d.; N=159 measurements) and 1.34±0.45 mg l^-1 in the hypoxic tanks(N=182 measurements). Temperature was 27±0.3°C, the salinity was 15.0±0.5 ppt (Fritz Super Salt Concentrate, Mesquite, TX,USA), and the photoperiod was 16 h:8 h light:dark.

Fish were fed to satiation with frozen brine shrimp (Artemia spp.,San Francisco Bay Brand, Newark, CA, USA), Prime Tropical Flakes (Ziegler Brothers, Inc., Gardners, PA, USA), and brine shrimp nauplii (O.S.I.,Snowville, UT, USA) twice daily. The diet was changed to live grass shrimp(Palaemonetes spp.) after 2 weeks. At the beginning of the exposure,fish in the two treatments were equivalent in standard length (normoxia,80.7±6.7 mm; hypoxia 79.9±4.6 mm) and mass (normoxia,10.8±3.2 g; hypoxia 10.0±2.3 g). Both groups grew over the 4 week exposure, although normoxic fish grew more than hypoxic fish (C. A. Landry, S. L. Steele, S. Manning and A. O. Cheek, manuscript submitted for publication). Consequently, normoxic fish were longer (normoxia,84.8±6.9 mm; hypoxia, 80.7±5.3 mm; P≤0.05) and heavier (normoxia, 15.2±3.6 g; hypoxia, 11.7±2.3 g; P≤0.01) than hypoxic fish at the end of the experiment.

After 4 weeks of exposure to normoxia or hypoxia, fish were netted and killed with an overdose of buffered MS-222 (1 g MS-222 and 4 g NaHCO₃ per liter of water). Liver, brain, heart and white skeletal muscle were dissected, frozen in liquid nitrogen, and stored at -80°C until analysis. Skeletal muscle samples were taken dorsal to the lateral line,to avoid red muscle, and between the head and dorsal fin, to avoid longitudinal variation in enzyme activity(Martínez et al.,2000).

For glycolytic and gluconeogenic enzyme assays, tissues were weighed and homogenized in ice-cold buffer consisting of 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 0.1 mmol l^-1 DTT and 0.2% Triton X-100 (Pierce and Crawford,1997) with a PRO 200 homogenizer (PRO Scientific Inc.,Connecticut, USA) for two 20 s periods. The samples were maintained on ice during and between periods of homogenization. Muscle, liver and brain samples were homogenized in nine volumes of buffer; hearts were homogenized in 49 volumes of buffer. Homogenates were centrifuged at 2400 g for 15 min at 4°C, and supernatant solutions were kept on ice until enzyme activity was assayed.

Separate homogenates were made for assays of glycogen synthase and glycogen phosphorylase. For these, liver and white muscle were weighed and homogenized in four volumes of ice-cold buffer containing 50 mmol l^-1imidazole, pH 7.5, 100 mmol l^-1 NaF, 5 mmol l^-1 EDTA, 5 mmol l^-1 EGTA, 15 mmol l^-1 β-mercaptoethanol and 0.1 mmol l^-1 phenylmethylsulfonyl fluoride (PMSF)(Milligan, 2003). Samples were homogenized with a PRO 200 homogenizer for two 20 s periods while kept cold. These homogenates were centrifuged at 16 000 g for 2 min at 4°C, and supernatants were kept on ice until enzyme activity was assayed.

Reaction conditions for the determination of glycolytic enzyme activities were modified from Pierce and Crawford(Pierce and Crawford, 1994). For each enzyme in each tissue, the concentrations of substrates, cofactors and linking enzymes were optimized to give maximal activities. Reactions were initiated by adding the substrate specific for that enzyme (shown last for each reaction). The final reaction conditions were as follows.

Hexokinase (HK; EC 2.7.1.1): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 7.5 mmol l^-1 MgCl₂,3.1 mmol l^-1 ATP, 1 mmol l^-1 NADP, 10 mmol l^-1 creatine phosphate, 2 i.u. ml^-1 creatine kinase, 1 i.u. ml^-1 glucose-6-phosphate dehydrogenase and 7.5 mmol l^-1 glucose. Under these conditions, glucokinase (hexokinase type IV) contributes to the rates measured in liver tissue.

Phosphoglucoisomerase (PGI; EC 5.3.1.9): 100 mmol l^-1Hepes (pH 7.4), 10 mmol l^-1 KCl, 1.25 mmol l^-1 NADP, 0.5 i.u. ml^-1 glucose-6-phosphate dehydrogenase, and 2 mmol l^-1 fructose 6-phosphate.

Phosphofructokinase (PFK; EC 2.7.1.11): 100 mmol l^-1Hepes (pH 8.2), 10 mmol l^-1 KCl, 7.5 mmol l^-1MgCl₂, 1.25 mmol l^-1 ATP (liver, brain, heart) or 2.5 mmol l^-1 ATP (muscle), 5 mmol l^-1 AMP, 0.2 mmol l^-1 NADH, 1 i.u. ml^-1 aldolase, 10 i.u. ml^-1glycerol-3-phosphate dehydrogenase, 29 i.u. ml^-1 triose phosphate isomerase and 5 mmol l^-1 fructose 6-phosphate (liver, brain, heart)or 10 mmol l^-1 fructose 6-phosphate (muscle).

Aldolase (ALD; EC 4.1.2.13): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 0.2 mmol l^-1 NADH, 5 i.u. ml^-1 glycerol-3-phosphate dehydrogenase, 14.5 i.u. ml^-1triose phosphate isomerase and 0.75 mmol l^-1 fructose 1,6-bisphosphate.

Triose phosphate isomerase (TPI; EC 5.3.1.1): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 0.2 mmol l^-1 NADH, 10 i.u. ml^-1 glycerol-3-phosphate dehydrogenase (muscle, liver, heart) or 20 i.u. ml^-1glycerol-3-phosphate dehydrogenase (brain) and 2.9 mmol l^-1glyceraldehyde 3-phosphate (muscle, liver, brain) or 5.8 mmol l^-1glyceraldehyde 3-phosphate (heart).

Glyceraldehyde-3-phosphate dehydrogenase (GAPDH; EC 1.2.1.12): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 2 mmol l^-1 MgCl₂ (muscle, brain, heart) or 1 mmol l^-1 MgCl₂ (liver), 3.1 mmol l^-1 ATP (muscle,brain, heart) or 1.55 mmol l^-1 ATP (liver), 0.2 mmol l^-1NADH, 8 i.u. ml^-1 phosphoglycerokinase and 2.8 mmol l^-13-phosphoglycerate.

Phosphoglycerokinase (PGK; EC 2.7.2.3): 100 mmol l^-1Hepes (pH 7.4), 10 mmol l^-1 KCl, 10 mmol l^-1MgCl₂, 3.1 mmol l^-1 ATP (liver, brain, heart) or 6.2 mmol l^-1 ATP (muscle), 0.2 mmol l^-1 NADH, 8 i.u. ml^-1 glyceraldehyde-3-phosphate dehydrogenase, and 2.8 mmol l^-1 3-phosphoglycerate.

Phosphoglyceromutase (PGM; EC 2.7.5.3): For liver, brain and heart, the assay included 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 5 mmol l^-1 MgCl₂, 0.65 mmol l^-1 ADP, 0.125 mmol l^-1 2,3-bisphosphoglycerate, 0.22 mmol l^-1 NADH, 9 mmol l^-1 glucose, 0.1 i.u. ml^-1 enolase, 0.5 i.u. ml^-1 pyruvate kinase, 0.75 i.u. ml^-1l-lactate dehydrogenase, 3.2 i.u. ml^-1hexokinase and 1.25 mmol l^-1 3-phosphoglycerate. For muscle, the above conditions were used except MgCl₂ was 2.5 mmol l^-1, ADP was 1.25 mmol l^-1, 2,3-bisphosphoglycerate was 62.5 μmol l^-1 and glucose was 5 mmol l^-1.

Enolase (ENO; EC 4.2.1.11): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 2.5 mmol l^-1 MgCl₂,1.3 mmol l^-1 ADP (muscle, liver, brain) or 0.65 mmol l^-1ADP (heart), 0.2 mmol l^-1 NADH, 4.5 mmol l^-1 glucose,0.6 i.u. ml^-1 pyruvate kinase, 0.75 i.u. ml^-1l-lactate dehydrogenase, 1.6 i.u. ml^-1 hexokinase(muscle, liver, brain) or 3.2 i.u. ml^-1 (heart) and 1.25 mmol l^-1 2-phosphoglycerate.

Pyruvate kinase (PYK; EC 2.7.1.40): 100 mmol l^-1 Hepes(pH 7.4), 10 mmol l^-1 KCl, 10 mmol l^-1 MgCl₂(muscle and liver) or 5 mmol l^-1 MgCl₂ (brain and heart), 7.6 mmol l^-1 ADP (muscle and liver) or 3.8 mmol l^-1 ADP (brain and heart), 0.2 mmol l^-1 NADH, 1.5 i.u. ml^-1l-lactate dehydrogenase (muscle), 0.375 i.u. ml^-1l-lactate dehydrogenase (liver and heart), or 0.75 i.u. ml^-1l-lactate dehydrogenase (brain), and 1 mmol l^-1 phosphoenolpyruvate (muscle, liver, brain) or 2 mmol l^-1 phosphoenolpyruvate (heart).

Lactate dehydrogenase (LDH; EC 1.1.1.27): 100 mmol l^-1Hepes (pH 7.4), 10 mmol l^-1 KCl, 0.17 mmol l^-1 NADH, 1 mmol l^-1 pyruvate.

The assay conditions for enzymes of glycogen metabolism were modified from Milligan (Milligan, 2003).

Glycogen synthase (GSase; EC 2.4.1.11): 50 mmol l^-1Tris (pH 7.8), 70 mmol l^-1 KCl, 4 mmol l^-1MgCl₂, 0.5 mmol l^-1 phosphoenolpyruvate, 0.2 mmol l^-1 NADH, 5 i.u. ml^-1l-lactate dehydrogenase, 5 i.u. ml^-1 pyruvate kinase, 2 mg ml^-1glycogen (oyster muscle, dialyzed) and 2 mmol l^-1 UDP-glucose. Total GSase activity was assayed in the presence of 5 mmol l^-1glucose 6-phosphate. Active GSase was measured without glucose 6-phosphate.

Glycogen phosphorylase (GPase; EC 2.4.1.1): 50 mmol l^-1potassium phosphate (pH 7.3), 15 mmol l^-1 MgSO₄, 0.5 mmol l^-1 DTT, 0.5 mmol l^-1 NADP, 0.25 mmol l^-1 EDTA, 1 i.u. ml^-1 glucose-6-phosphate dehydrogenase,1 i.u. ml^-1 phosphoglucomutase, 0.01 mmol l^-1 glucose 1,6-bisphosphate and 2 mg ml^-1 glycogen (oyster muscle, dialyzed). Total GPase activity was measured in the presence of 2 mmol l^-1AMP. Active GPase was measured without AMP.

The assay conditions for gluconeogenic enzymes were modified from standard protocols.

Malate dehydrogenase (MDH; EC 1.1.1.37): 100 mmol l^-1Hepes (pH 7.4), 10 mmol l^-1 KCl, 0.175 mmol l^-1 NADH and 0.1 mmol l^-1 oxaloacetate(Mommsen et al., 1980).

Phosphoenolpyruvate carboxykinase (PEPCK; EC 4.1.1.32): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 10 mmol l^-1 phosphoenolpyruvate, 0.5 mmol l^-1 inosine diphosphate, 5 mmol l^-1 MnCl₂, 0.15 mmol l^-1NADH, 0.3 i.u. ml^-1 malate dehydrogenase, and 20 mmol l^-1 NaHCO₃ (Opie and Newsholme, 1967). In optimizing the PEPCK assay, inosine diphosphate (IDP) and 2-deoxy-guanosine-5′-phosphate (2-dGDP)(Foster and Moon, 1990) were compared as phosphoryl group acceptors. With IDP, background rates (without NaHCO₃) were lower and specific rates (with NaHCO₃) were higher than with 2-dGDP. The resulting PEPCK activities were as much as 50%greater with IDP as the phosphoryl group acceptor. The greater activity cannot be due to competing pyruvate kinase activity, because the PEPCK assay is initiated with NaHCO₃.

Fructose-1,6-bisphosphatase (FBPase; EC 3.1.3.11): 100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl, 2 mmol l^-1 MgCl₂, 1 mmol l^-1 EDTA, 0.2 mmol l^-1 NADP, 1.6 i.u. ml^-1 phosphoglucose isomerase, 0.36 i.u. ml^-1 glucose-6-phosphate dehydrogenase, and 0.05 mmol l^-1 fructose 1,6-bisphosphate(Opie and Newsholme,1967).

Glucose-6-phosphatase (G6Pase; EC3.1.3.9): 100 mmol l^-1Hepes (pH 6.5), 10 mmol l^-1 KCl, 26.5 mmol l^-1 glucose 6-phosphate, 1.8 mmol l^-1 EDTA, 2 mmol l^-1 NAD, 1 i.u. ml^-1 mutarotase, 20 i.u. ml^-1 glucose dehydrogenase(Alegre et al., 1988). This assay was initiated by the addition of extract.

Maximal enzyme activities were measured in quadruplicate in a 96-well microplate reading spectrophotometer (VERSAmax, Molecular Devices, Sunnyvale,CA, USA) at 27±1°C. Reaction rates were linear for ≥3 min. Rates from blank reactions (without substrate) were subtracted for all determinations of enzyme activities. Units (i.u.) of enzyme activity were defined as the amount of enzyme needed to convert 1 μmol of substrate to product in 1 min under these conditions. The value of 6.22 was used as the millimolar extinction coefficient for NAD(P)H. All enzyme activities were measured within 5 h of tissue homogenization.

Biochemicals and coupling enzymes were purchased from Sigma Chemical Co.(St Louis, MO, USA), Roche Diagnostics Corporation (Indianapolis, IN, USA) or Calzyme Laboratories, Inc. (San Luis Obispo, CA, USA). When necessary to remove excess ammonium sulfate, coupling enzymes were centrifuged at 12,000 g for 10 min and redissolved in assay buffer [100 mmol l^-1 Hepes (pH 7.4), 10 mmol l^-1 KCl].

The protein contents in the supernatant fractions of tissue homogenates were determined by the bicinchoninic acid assay(Smith et al., 1985; Brown et al., 1989), modified for use in a 96-well microplate reading spectrophotometer. Samples were diluted in water to a concentration of approximately 0.5 mg ml^-1. Quadruplicate 10 μl samples were added to 200 μl of the bicinchoninic acid working reagent (Pierce Biochemicals, Rockford, IL, USA) in individual wells of a 96-well microplate. Wells were sealed and the plate was incubated at 60°C for 30 min. After cooling to room temperature, the plate was read at 562 nm. Standards of 0-1 mg ml^-1 bovine serum albumin were included with every plate.

Enzyme activities were calculated on the basis of tissue mass (i.u. g^-1 tissue) and on the basis of soluble protein content in tissue extracts (i.u. mg^-1 protein). Because the same fish were used to measure the effects of low oxygen on reproduction (C. A. Landry, S. L. Steele,S. Manning and A. O. Cheek, manuscript submitted for publication), the effects of DO treatment on enzyme activities were evaluated with 2-way analyses of variance which included the sex of the fish and the interaction between DO and sex. In this model, a significant effect of sex (i.e. enzyme activities in females differ from males) or a significant interaction (i.e. the effect of DO depended upon sex of the fish) would be taken as evidence that reproductive status affects enzyme activity. Throughout, enzyme activities are presented as least-squared means for normoxic and hypoxic treatments (corrected for the effects of sex and the interaction between sex and DO) with one standard deviation (s.d.). We present statistical results at two levels of probability:one that assumes independence among variables (P≤0.05); and one that allows for enzyme responses within a pathway and across tissues to be linked (P≤0.001). The latter approach corrects P values for multiple comparisons (Sokal and Rohlf,1981). All statistical analyses were performed with SYSTAT 10.

Maximal activities of enzymes of glycolysis, glycogen metabolism and gluconeogenesis were measured in tissues of F. grandis after 4 weeks of exposure to normoxia or hypoxia. Enzyme activities expressed per gram of tissue are presented in Tables S1-S3 in supplementary material. Expressed in this way, differences in enzyme activity between normoxic and hypoxic fish might be explained by treatment-related changes in tissue protein content. For example, the soluble protein concentration in skeletal muscle extracts from hypoxic fish (40.7±5.2 mg g^-1 tissue) was significantly lower than in skeletal muscle extracts from normoxic fish (48.4±5.8 mg g^-1 tissue; P≤0.01). Similarly, the soluble protein concentration was lower in heart from hypoxic fish (71.7±8.9 mg g^-1 tissue) than from normoxic fish (81.6±13.6 mg g^-1 tissue; P≤0.05). [The soluble protein concentrations did not differ between normoxic and hypoxic treatments for extracts prepared from liver (normoxia= 108.7±11.1 mg g^-1tissue; hypoxia=109.2±14.9 mg g^-1 tissue) or brain(normoxia=60.6±6.9 mg g^-1 tissue; hypoxia= 60.5±6.6 mg g^-1 tissue).] If enzyme activities in the two treatment groups simply paralleled the concentration of soluble protein, then activities per gram of tissue mass would be 10% (heart) to 20% (skeletal muscle) lower in hypoxia than in normoxia. To account for treatment effects on tissue protein concentration, enzyme specific activities (in i.u. mg^-1 protein)were calculated. These values reflect variation beyond that due to changes in bulk protein levels, and these values were used for the remainder of the analyses.

Hypoxia altered glycolytic enzyme activities in white skeletal muscle,liver, heart, and brain; however, the enzymes affected and the direction of the hypoxia response were tissue-specific(Table 1; Fig. 1). Hypoxic exposure led to significantly lower specific activities of eight of ten glycolytic enzymes measured in white skeletal muscle (P≤0.05; Fig. 1A). Activities were reduced by 17% (PGM) to 55% (ENO). The two other enzymes (GAPDH and PGK) were lower in hypoxia than in normoxia, although these differences were not statistically significant. The enzyme HK was below the limit of detection in skeletal muscle. In contrast to the effect of hypoxia in skeletal muscle,hypoxic exposure enhanced activities of five of 11 glycolytic enzymes measured in liver (P≤0.05; Fig. 1B). Activities were 19% (PGI) to 74% (PGK) greater in hypoxia. In heart, three glycolytic enzymes had higher specific activities in hypoxia-exposed fish (P≤0.05; Fig. 1C), ranging from 18%(TPI) to 28% (HK) increases. In brain, the effects of hypoxia on maximal enzyme activities were smaller in magnitude and mixed in direction: three glycolytic enzymes had higher activities in brains from hypoxia-exposed fish,whereas one had lower activity (P≤0.05; Fig. 1D). The percentage changes ranged from 12% lower (PGM) to 16% higher (HK) in hypoxia.

White skeletal muscle and liver had undetectable or low levels of hexokinase, suggesting that free glucose is less important than stored glycogen as a substrate for glycolysis. Accordingly, to get a more complete picture of overall carbohydrate metabolism, the total and active levels of the enzymes of glycogen metabolism, GPase and GSase, were determined in skeletal muscle and liver (Table 2; Fig. 2). Both total and active GPase activities were lower in skeletal muscle from hypoxia-exposed fish(P≤0.05; Fig. 2A). This reduction in enzyme activity of glycogenolysis is consistent with lower activities of glycolytic enzymes in hypoxic muscle (see above). In skeletal muscle, the percentage of GPase in the active form was about 15% of the total GPase activity and did not differ between normoxia and hypoxia. Dissolved oxygen treatment had no effect on the specific activity of GPase (either total or active) in liver (Fig. 2B). However, small, non-statistically significant changes in liver GPase specific activity led to a modest, but significant decrease in the percentage of active GPase in liver in hypoxic fish (73±11%) compared to normoxic fish(83±10%) (P≤0.05). Total GSase activity in muscle was lower in hypoxia (P≤0.05; Fig. 2A). The same enzyme was significantly higher in liver tissue from hypoxia-exposed fish (Fig. 2B),although this was due to an effect of hypoxia on males but not females (see below). Active GSase was not significantly altered by hypoxia in either tissue, nor was the percentage of GSase in the active form (12-15% in both tissues).

The specific activities of enzymes that are involved in gluconeogenesis were measured only in liver (Table 3; Fig. 3). The citric acid cycle enzyme MDH catalyzes the reversible conversion of oxaloacetate to malate, which may be important during gluconeogenesis as a mechanism to shuttle reducing equivalents and carbon skeletons from the mitochondrion to the cytosol. This enzyme was significantly higher in hypoxia-exposed fish (P≤0.05). The enzyme FBPase catalyzes the conversion of fructose 1,6-bisphosphate to fructose 6-phosphate, bypassing the glycolytic reaction catalyzed by PFK, and it was also higher in hypoxic fish(P≤0.05). Both enzymes were about 25% higher in hypoxia. Two other enzymes of gluconeogenesis, PEPCK and G6Pase, did not differ between normoxic and hypoxic fish.

The specific activities of six enzymes differed between male and female fish (P≤0.05). Of these, five [ALD in heart, and MDH, FBPase,GPase (total), and GPase (active) in liver] were higher in males. These differences were relatively modest, generally being less than 25%. By contrast, the only enzyme which was greater in females (liver PEPCK) was nearly 80% greater than in males. For one enzyme (total liver GSase), there was a significant DO by sex interaction (P≤0.05). Males and females had equivalent activities under normoxia; hypoxic exposure led to higher levels in males, but not in females.

While a P value of less than 0.05 is typically accepted as demonstrating significant differences between treatment groups, using this value assumes that the measured variables are independent of one another. It is possible that changes in enzyme activities in a given metabolic pathway occur in a coordinated fashion (i.e. they are not independent of one another),therefore the P value must take into account the total number of enzyme activities measured (approximately 50 different enzyme-tissue combinations). Even using this more stringent criterion (P≤0.001),a number of enzyme activities were found to differ between normoxic and hypoxic fish. Moreover, this approach clearly demonstrates that the response to hypoxia differed among the four tissues(Table 4). The only effect of sex that was significant at P≤0.001 was the higher value of liver PEPCK in females.

We undertook this study to develop a comprehensive picture of carbohydrate metabolism in multiple tissues from a single fish species subjected to prolonged hypoxia. We measured maximal activities of enzymes of glycolysis,glycogen metabolism, and gluconeogenesis in white skeletal muscle, liver,heart and brain of hypoxic and normoxic F. grandis. These measurements allowed us to address the following questions: do different tissues respond similarly to hypoxic exposure?; do all enzymes within a given pathway change in concert? And, do enzymes in catabolic and anabolic processes respond similarly or differently to chronic hypoxia? In addressing these questions, we concentrate on those results that satisfy the criterion of being statistically significant after accounting for multiple comparisons(Table 4). Moreover, the statistical analyses included sex of the fish, so the observed effects of DO are independent of any sex-dependent effects (e.g. changes in reproductive status) that might have occurred during the acclimation period.

In F. grandis, we found that the specific activities of enzymes of glycolysis and glycogen metabolism were consistently depressed after long-term exposure to low oxygen, suggesting a decreased capacity for carbohydrate metabolism in this tissue. The lower enzyme activities may be related to a reduction in growth or activity during chronic hypoxia. Hypoxia leads to lower growth rates in this and other teleost species(Chabot and Dutil, 1999; Stierhoff et al., 2003; C. A. Landry, S. L. Steele, S. Manning and A. O. Cheek, manuscript submitted for publication), and fish experiencing low or negative growth frequently lose skeletal muscle protein, presumably due to catabolism to meet energy demands(Sullivan and Somero, 1983; Loughna and Goldspink, 1984; Pelletier et al., 1993; Pelletier et al., 1995; Martínez et al., 2003). This is consistent with our observation that soluble protein concentrations(mg g^-1 tissue mass) were about 20% lower in skeletal muscle extracts from hypoxic fish compared to normoxic fish. If glycolytic enzymes were catabolized at the same rate as bulk proteins, then enzyme activities would decrease by a similar amount (20%) when expressed on the basis of tissue mass. Instead, skeletal muscle enzyme activities per gram tissue decreased by more than this (see Tables S1-S3 in supplementary material). Consequently,enzyme-specific activities (which account for changes in protein content) were also lower in hypoxia relative to normoxia (Tables 1, 2), demonstrating that the changes we noted were above those due to loss of bulk protein associated with decreased growth. Furthermore, we included growth rate of individual fish in preliminary statistical analyses of enzyme activity data (not shown). In general, the effects of growth rate were small and not consistent in direction: in skeletal muscle, one enzyme specific activity was positively related to growth rate and one was negatively related to growth rate. More importantly, the conclusion that skeletal muscle enzyme activities were significantly affected by hypoxia was not changed even when growth rate was included in the analyses.

Therefore, a second explanation for the reduced muscle enzyme activities relates to decreased locomotion during hypoxia. Behavioral observations in this and other studies have shown hypoxic fish to be less active than normoxic fish (Bushnell et al., 1984; Dalla Via et al., 1998; Wannamaker and Rice, 2000),and, over the long term, the lower energy demands associated with decreased locomotion might result in lower levels of glycolytic enzymes. A similar link between muscle metabolic and locomotory capacities has been forwarded to account for the observed inter- and intraspecific variation in glycolytic enzyme activities in fish of differing size, lifestyle or condition(Somero and Childress, 1980; Childress and Somero, 1990; Martínez et al.,2003).

In contrast to skeletal muscle, hypoxic acclimation of F. grandisled to increased specific activities of enzymes of carbohydrate metabolism in liver. These results suggest that chronic hypoxia leads to an increase in the capacity for carbohydrate metabolism in this tissue. Perhaps surprisingly, we measured higher activities for enzymes of carbohydrate catabolism (glycolysis)and carbohydrate anabolism (glycogen synthesis and gluconeogenesis). It seems paradoxical that enzymes of both catabolic and anabolic pathways were higher under hypoxia. Although certain enzymes are shared between glycolysis and gluconeogenesis (GAPDH, PGK, LDH), other enzymes are specific to catabolic or anabolic reactions (i.e. PFK in glycolysis versus FBPase in gluconeogenesis). Increased activities of the latter enzymes suggest a futile cycle whose net result is ATP turnover. However, there is evidence that teleost liver contains distinct cell populations that differ from one another in their glycolytic and gluconeogenic capacities(Mommsen et al., 1991), a functional separation that would have been lost during homogenization. Therefore, it is possible that glycolysis could be upregulated in one population of cells to enhance ATP production to meet the energetic demands of those cells while gluconeogenesis could be upregulated in another cell population to enhance endogenous glycogen synthesis or export of glucose to extra-hepatic tissues. Amino acids coming from the catabolism of skeletal muscle protein during hypoxia could serve as precursors for this increased gluconeogenic flux.

In heart and brain, fewer enzyme activities differed between normoxic and hypoxic F. grandis than in other tissues, and these differences were generally smaller in magnitude. One enzyme that did change in both tissues was hexokinase, which was higher in heart and brain from fish subjected to chronic hypoxia. Because the rate of glucose utilization by teleost brain is thought to be limited by hexokinase activity(Soengas and Aldegunde, 2002),the greater hexokinase activity could result in higher rates of glycolysis in this tissue during hypoxia. Otherwise, the relatively subtle changes in heart and brain might be explained by these tissues receiving preferential blood flow, and hence oxygen delivery, during exposure of fish to low oxygen(Soengas and Aldegunde, 2002). In accord with our results, heart and brain showed fewer signs of metabolic imbalance than skeletal muscle and liver during acute exposure of rainbow trout to hypoxia (Dunn and Hochachka,1986). Of course, all of the tissues studied may respond to low oxygen by other mechanisms that would not be reflected as changes in enzyme specific activities. In this regard, goldfish exposed to anoxia respond with large increases in the concentration of fructose 2,6-bisphosphate, a potent allosteric activator of PFK, in heart and brain tissues(Storey, 1987).

In no tissue did all enzymes of the glycolytic pathway change, at least in a statistically significant fashion. Among the tissues examined, skeletal muscle was characterized by the most consistent changes: eight of 10 enzymes were significantly lower in hypoxia at P≤0.05; four were significant at P≤0.001. In liver, heart and brain, the number of significant changes depended upon the P value and the tissue, but ranged from a minimum of one enzyme to a maximum of five (fewer than half of the 11 glycolytic enzymes measured in these tissues). In all tissues, hypoxic exposure affected one enzyme typically considered to be `rate-limiting' for glycolysis: PYK in muscle, PFK in liver and HK in heart and brain. However,hypoxia also affected the activities of several `near-equilibrium' enzymes in muscle and liver. These results suggest that the designation `rate-limiting'or `near-equilibrium' is not a reliable predictor of which enzymes might be affected by a particular experimental treatment. Our results support the conclusions that biologically meaningful variation in enzyme activity occurs in reactions not usually thought to be rate-determining for the glycolytic pathway (Pierce and Crawford,1997; Kraemer and Schulte,2004).

It has been suggested that glycolytic enzymes in a variety of organisms are coordinately upregulated during hypoxic stress(Webster, 2003). Our data do not support a simple interpretation of this hypothesis, which predicts increased levels of all glycolytic enzymes in a given tissue during hypoxia. Indeed, for the tissue with the most consistent changes (skeletal muscle), the changes were in a direction opposite of that predicted. Two possible explanations of why our conclusions differ from those predicted are that the efficiency of oxygen extraction by fish increased, or demands for energy production by tissues decreased, during long-term hypoxic exposure. In other species of fish, the capacity to extract oxygen from hypoxic waters increases during chronic hypoxic exposure, through adjustments in ventilation, oxygen transport or gill surface area (Jensen et al., 1993; Sollid et al.,2003). The result is higher rates of oxygen consumption at low oxygen (Johnston and Bernard,1982) or a decrease in critical oxygen tension(Timmerman and Chapman, 2004). An increase in oxygen extraction by F. grandis would presumably lessen the need for `compensatory' changes in anaerobic capacity as the duration of hypoxia is extended. With respect to energy demands, quantitative estimates of metabolism in fish held under hypoxia suggest that the increase in anaerobic metabolism is smaller than that expected from the decrease in aerobic metabolism (Dalla Via et al.,1994; Virani and Rees,2000). This appears to be true in F. grandis, where overall metabolism (aerobic plus anaerobic components) is lower during exposure to hypoxia than normoxia (Virani and Rees, 2000). Indeed, metabolic rate reduction has been proposed as a key feature enabling hypoxic survival in fish and other hypoxia-tolerant animals (Hochachka,1980; Hochachka and Somero,2002). The overall result is that the tissue response to chronic hypoxia is heterogeneous, and it reflects the interplay among energetic demands, metabolic role, and oxygen supply of specific tissues. Thus, the paradigm of uniformly increased glycolytic enzyme potential might describe the response of a particular tissue at a specific duration of hypoxia, but it may not be the appropriate solution for the long-term response of fish to low oxygen.

List of abbreviations
 
• ALD

  aldolase

 
• DO

  dissolved oxygen

 
• 2-dGDP

  2-deoxy-guanosine-5′-phosphate

 
• ENO

  enolase

 
• FBPase

  fructose-1-6-bisphosphatase

 
• GAPDH

  glyceraldehyde-3-phosphate dehydrogenase

 
• GPase

  glycogen phosphorylase

 
• G6Pase

  glucose-6-phosphatase

 
• GSase

  glycogen synthase

 
• Hepes

  N-[2-hydroxyethyl]piperazine-N′-[2-ethanesulfonic acid]

 
• HK

  hexokinase

 
• IDP

  inosine diphosphate

 
• LDH

  lactate dehydrogenase

 
• MDH

  malate dehydrogenase

 
• MS-222

  ethyl 3-aminobenzoate methane-sulfonate salt

 
• PEPCK

  phosphoenolpyruvate carboxykinase

 
• PFK

  phosphofructokinase

 
• PGI

  phosphoglucoisomerase

 
• PGK

  phosphoglycerokinase

 
• PGM

  phosphoglyceromutase

 
• PYK

  pyruvate kinase

 
• TPI

  triose phosphate isomerase

We are grateful to Stacy Steele for assistance during sampling and to Deb Vivian for advice on fish culture. This project was funded by NSF grant No. IBN 0236494 to B.B.R. and US-EPA STAR Program grant No. R829458 to A.O.C.