Pathways for urea production during early life of an air-breathing teleost, the African catfish Clarias gariepinus Burchell

Urea rarely constitutes more than 10–20% of the total nitrogen excreted by teleosts (Wood, 1993), except in a few species such as the Lake Magadi tilapia Oreochromis alcalicus grahami (Randall et al., 1989). In early teleost life stages, however, urea may be a more important end-product of nitrogen metabolism, as shown for Zoarces viviparous (Korsgaard, 1994) and the Atlantic cod (Gadus morhua, Chadwick and Wright, 1999). During early development of the African catfish Clarias gariepinus, urea constitutes 62% of the total nitrogen excreted (as ammonia-N + urea-N) in embryos, 20% during yolk-sac and starved larval stages, and 44% after metamorphosis in fed larvae (Terjesen et al., 1997). In contrast, immersed adult C. gariepinus (Eddy et al., 1980) and C. batrachus (Saha and Ratha, 1989) excrete less urea, accounting for 13% and 15% of total nitrogen excreted, respectively. The African catfish is an air-breathing teleost tolerant to wide fluctuations in water availability and temperature (Donnely, 1973). Spawning occurs shortly after heavy rainfall, and the embryos are deposited in a few centimetres of water on vegetation in temporarily flooded areas (Greenwood, 1955; Bruton, 1979).

Three pathways may be responsible for urea production in teleosts: hydrolysis of arginine (argininolysis), catabolism of uric acid (uricolysis) and the ornithine–urea cycle (OUC). Arginine from the diet or from protein turnover is hydrolyzed by the enzyme arginase (Wood, 1993). On the basis of arginine depletion rates in C. gariepinus yolk-sac and starved larvae in vivo, Terjesen et al. (Terjesen et al., 1997) showed that argininolysis could account for approximately one-third of the urea excreted, suggesting that other pathway(s) for urea synthesis were also functional.

In adult teleosts, uricolysis serves to degrade purines originating from the diet or from nucleic acid turnover and may be the dominant pathway for urea synthesis in adults of most species (Goldstein and Forster, 1965; Vellas and Serfaty, 1974; Wright, 1993). During early development, de novo purine synthesis can be expected to provide building blocks for larval growth. Given that nucleotides are energetically expensive to synthesize (Stryer, 1988), it is probable that a rapidly growing yolk-sac larva that relies on endogenous nutrition would not express enzymes for purine degradation. However, to our knowledge, no studies have investigated the uricolytic pathway in teleost embryos and larvae.

In teleosts, urea may also be formed from glutamine, HCO₃⁻¹, ATP and aspartate by the OUC (Anderson, 1995). Only a few teleost species express significant activities of OUC enzymes in the adult (e.g. Mommsen and Walsh, 1989; Randall et al., 1989; Saha and Ratha, 1989), whereas all species so far studied express OUC enzymes during early life, as reported for guppy (Poecilia reticulata) (Dépêche et al., 1979), rainbow trout (Oncorhynchus mykiss) (Wright et al., 1995), Atlantic cod (Gadus morhua) (Chadwick and Wright, 1999) and Atlantic halibut (Hippoglossus hippoglossus) (Terjesen et al., 2000). Expression of the OUC in early teleost life stages has been hypothesized to assist in detoxification of ammonia accumulating from amino acid catabolism, which may be a particular problem for species with large embryos (Griffith, 1991; Wright et al., 1995). The early life stages of C. gariepinus are relatively small, however (individuals weigh 1–2.5mg), and, although ammonia levels also increase in yolk-sac larvae of C. gariepinus, considerably more ammonia is excreted than is accumulated (Terjesen et al., 1997).

The main objective of the present study was to describe the biochemical pathways that may be responsible for the high rate of urea excretion by C. gariepinus larvae (Terjesen et al., 1997). We measured the activities of the enzymes of uricolysis, argininolysis and the OUC in the early life stages of C. gariepinus and in adult C. gariepinus tissues to investigate whether developmental stage influences the expression of the urea-producing pathways. If enzymes for the OUC were expressed, another aim of this study was to describe the basic kinetics and tissue location of the key regulatory enzyme carbamoyl phosphate synthetase III (CPSase III). In addition, we measured urea and uric acid levels in early life stages, which adds to data from our previous study on urea and ammonia excretion and tissue ammonia concentrations (Terjesen et al., 1997), to provide a detailed view of nitrogen metabolism and excretion in this species.

This study was approved by the Animal Experiment Committee at Wageningen University. Fertilized embryos of Clarias gariepinus Burchell were produced by artificial reproduction, as outlined previously (Hogendoorn and Vismans, 1980; Terjesen et al., 1997). Water temperature was maintained at 27.8±0.2°C, pH8.3±0.1, and electrical conductivity at 295±60μScm⁻¹ (means ± s.d.). Total ammonia levels in the aquarium outlet water did not exceed 5μmoll⁻¹ (Orion 9512 electrode, Thermo Orion, MA, USA). A fed group (fed larvae) was offered live Artemia franciscana instar I nauplii (from RH grade cysts, INVE Aquaculture, Belgium; according to the method of Verreth and Den Bieman, 1987) from 73h post-fertilization. An additional group (starved larvae) was included to study the effects of starvation (terminated at 194h post-fertilization).

Sampling was conducted according to the method of Terjesen et al. (Terjesen et al., 1997); in particular, fed larvae were not collected until 12h after the last feeding. Samples were taken in triplicate for wet and dry mass and urea determination by collecting 2–20 individuals. Because of limited tissue availability, only a few samples were collected for uric acid analysis.

Samples were taken in triplicate for both OUC and uricolytic enzyme studies by collecting 0.4–1.2g of embryos or larvae per sample according to the protocol of Terjesen et al. (Terjesen et al., 2000), followed by freezing in liquid nitrogen. All samples except those used for dry mass measurements were stored at −139°C during the experimental period in Wageningen. Thereafter, samples were transported to the University of Bergen and stored at −80°C until analysis.

For wet mass, dry mass and urea determinations, 10–20 yolk-sac larvae aged 30, 55 or 82h post-fertilization were dissected into yolk and body compartments (as described by Terjesen et al., 1997). The body compartment (i.e. yolk-free animal) is termed ‘larval body’. An attempt was made to establish the distribution of CPSase III in fed larvae aged 217h post-fertilization (30±2mgwetmass per individual). Livers (constituting less than 10% of body mass) were dissected from 100 individuals while they were held in ice-water. Samples of posterior body segments were collected to demonstrate possible expression of CPSase III in skeletal muscle tissue. Approximately 60 larvae were cut quickly in two (1–2mm behind the anus to avoid obtaining kidney tissue) while held in ice-water. Except for a minor contribution from bone, connective and neural tissue, the posterior segment should constitute mostly larval muscle tissue.

Tissue samples of adult C. gariepinus were collected for enzyme activity determinations, relative water content and urea measurements. One female (1.0kg) and two males (2.8 and 3.0kg), which had been fasted for 48h, were quickly killed by a sharp blow to the head, and blood samples were collected from posterior haemal arches. Thereafter, liver, kidney and posterior dorsal muscle tissues were excised, rinsed in 0.9% NaCl, blotted on tissue paper, weighed and then frozen in liquid nitrogen. Blood samples were centrifuged at 2240g for 10min, and plasma was frozen in liquid nitrogen. Samples were subsequently stored as described above.

All samples were processed at the University of Bergen in a cold-room at 4°C. Since uricolytic enzymes have not been measured in embryonic teleosts, variations in extraction and assay procedures were tested. The optimal extraction procedure was as follows. Frozen samples (0.2–1.1g) were homogenized in 2.5–6.0ml of fresh extraction buffer (50mmoll⁻¹ Hepes, pH7.5, 50mmoll⁻¹ KCl, 0.5mmoll⁻¹ EDTA, 5% glycerol, w/w, and 0.015mgml⁻¹ trypsin inhibitor) using an IKA T25 Basic Ultra-Turrax homogenizor with an S25N-10G knife at 8000revsmin⁻¹ for 12s and 24000revsmin⁻¹ for 12s. While incubating the tubes in ice-water, samples were sonicated for 1min, followed by a pause of approximately 4min, and then sonicated for 1min at an amplitude setting of 40 (Soncis and Materials Vibra-Cell, CV18 probe). Allantoicase and ureidoglycollate lyase (UGL) assays did not produce reliable results if shorter sonication times were employed. For uricase and allantoinase, however, no consistent significant differences between procedures were observed, so all the data were pooled. OUC enzymes were extracted according to the method of Terjesen et al. (Terjesen et al., 2000). After centrifugation for 10min at 14500g and 4°C, supernatants were passed through Sephadex G-25 columns (2cm×16cm; Pharmacia Biotech, Uppsala, Sweden) equilibrated with extraction buffer (as described by Terjesen et al., 2000).

As outlined by Anderson (Anderson, 1995), useful criteria for the detection of CPSase III in tissues also containing CPSase II (pyrimidine synthesis) are (i) activation by N-acetyl-l-glutamate (AGA), (ii) higher activity with glutamine than with ammonia as a substrate, (iii) lack of inhibition by uridine triphosphate (UTP) and (iv) separation of CPSase II and III by gel filtration chromatography and elution of a potential CPSase III peak at a position corresponding to a molecular mass of 160kDa. Thus, gel filtration chromatography of whole larvae (315h post-fertilization) and dissected liver and muscle extracts (217h post-fertilization) was conducted at 4°C (as described by Anderson and Walsh, 1995; Terjesen et al., 2000) on a Sephacryl S300HR column (1.6cm×60cm; Pharmacia Biotech) equilibrated with freshly prepared buffer containing 100mmoll⁻¹ KCl, 50mmoll⁻¹ Hepes (pH7.5), 0.5mmoll⁻¹ EDTA, 15mmoll⁻¹ MgCl₂, 10mmoll⁻¹ ATP, 10mmoll⁻¹ NaHCO₃, 2mmoll⁻¹ dithiothreitol, 0.015mgml⁻¹ trypsin inhibitor and 10% glycerol (w/w). CPSase elution profiles were constructed from assays with both glutamine and AGA present in the reaction mixture. The molecular masses of the eluted proteins were estimated using the method of Terjesen et al. (Terjesen et al., 2000). Protein concentration was determined after Bradford (Bradford, 1976).

Because they are markedly unstable, substrates, cofactors and standards were prepared shortly before use and incubated on ice. Each sample was run in duplicate, and controls containing no substrate or no extract were included. For each assay, standards were run with known concentrations of products or substrates dissolved in the same solution as the samples. Activity was calculated from the linear part of the curve (product formation or substrate depletion versus time) after subtracting the control rate. All reactions were performed at 26°C and were linear with respect to extract concentration.

Uricase (EC 1.7.3.3) was assayed essentially as described by Brown et al. (Brown et al., 1966). The final reaction mixture contained 60μmoll⁻¹ uric acid, 367mmoll⁻¹ glycine (pH9.7), 8.3mmoll⁻¹ Hepes (pH7.5), 8.3mmoll⁻¹ KCl, 0.1mmoll⁻¹ EDTA, 0.8% glycerol (w/w) and 0.0025mgml⁻¹ trypsin inhibitor in a final volume (i.e. including extract) of 1200μl. Before each assay, the reaction mixture was bubbled with O₂ at room temperature (20–25°C) for 60min. The reaction was initiated by the addition of 200μl of extract (10μl for adult liver) and followed by the decline in absorbance at 293nm (A₂₉₃) as uric acid was converted into the non-absorbing allantoin.

Allantoinase (EC 3.5.2.5.) was assayed essentially as described by Takada and Noguchi (Takada and Noguchi, 1983) employing differential analysis of glyoxylate derivatives (Vogels and Van der Drift, 1970). The final reaction mixture contained, in a volume of 500μl: 50mmoll⁻¹ Tris (pH8.2), 25mmoll⁻¹ allantoin, 12mmoll⁻¹ Hepes (pH7.5), 12mmoll⁻¹ KCl, 0.1mmoll⁻¹ EDTA, 1.2% glycerol (w/w) and 0.004mgml⁻¹ trypsin inhibitor. Reactions were initiated by addition of 120μl of extract (5μl for adult liver) and terminated at 0 and 60min by addition of 500μl of 0.5moll⁻¹ HCl, followed by boiling for 5min. Subsequently, the tubes were cooled on ice, and the contents were neutralized with 500μl of 0.5moll⁻¹ KOH in 0.1moll⁻¹ Hepes/0.1moll⁻¹ KCl. After centrifugation at 9600g, glyoxylate in the supernatant was determined by adding NADH and lactate dehydrogenase (LDH, Sigma L 2625) at 340nm. Assuming a NADH extinction coefficient of 6.22mmoll⁻¹cm⁻¹, standards gave a recovery of 74±18% (mean ± s.d., N=48). The incomplete recovery reduced the sensitivity of the assay but did not affect calculations since standards were included at each run.

Allantoicase (EC 3.5.3.4) protocols described in the literature showed considerable differences. Initially, the assay described by Takada and Noguchi (Takada and Noguchi, 1986) (procedure 1) was employed. This gave acceptable results when using adult mackerel (Scomber scombrus) liver as a positive control, but allantoicase activity in larval C. gariepinus extracts was not detectable using this assay. Thereafter, an assay employing urease (Sigma U 4002) with subsequent ammonia determination was developed, but reproducible results were not obtained using this method either. A continuous assay (after Streamer, 1980) based on that of Brown et al. (Brown et al., 1966) gave acceptable results, however. Variations on this protocol have been employed previously (Wright, 1993). The assay makes use of the fact that LDH will utilize glyoxylate as a substrate and concurrently oxidize NADH. The final reaction mixture contained 50mmoll⁻¹ Hepes (pH7.5), 50mmoll⁻¹ allantoate, 50mmoll⁻¹ KCl, 0.1mmoll⁻¹ EDTA, 0.7% glycerol (w/w), 0.002mgml⁻¹ trypsin inhibitor, 0.15mmoll⁻¹ NADH and 22unitsml⁻¹ LDH (Boehringer-Mannheim 127 221) in a volume of 720μl. Reactions were initiated by addition of 95μl of extract (30μl for adult liver). Glyoxylate standards gave a recovery of 99±12% (mean ± s.d., N=54). Activity was observed as a decline in A_340. This assay will not give correct results unless a higher activity of ureidoglycollate lyase (UGL) than of allantoicase is present, since ureidoglycollate and one urea molecule are formed in the allantoicase reaction while glyoxylate is formed only by UGL (Takada and Noguchi, 1986). However, in the present study, UGL showed considerably higher activity than allantoicase in all samples, and the use of this assay is therefore justified.

Ureidoglycollate lyase (UGL, EC 4.3.2.3) was assayed essentially as described by Pineda et al. (Pineda et al., 1994). The final reaction mixture contained 94mmoll⁻¹ Tes (pH7.8), 5mmoll⁻¹ Hepes (pH7.5), 5mmoll⁻¹ KCl, 2.7mmoll⁻¹ ureidoglycollate, 0.04mmoll⁻¹ EDTA, 0.4% glycerol (w/w), 0.002mgml⁻¹ trypsin inhibitor, 0.5mmoll⁻¹ MnCl₂, 0.25mmoll⁻¹ NADH and 26unitsml⁻¹ LDH (Boehringer-Mannheim 127 221) in a final volume of 799μl. Reactions were initiated by addition of 60μl of extract (10μl for adult liver) and followed as the decline in A₃₄₀. Glyoxylate standards gave a recovery of 99±8% (mean ± s.d., N=57).

Preliminary tests showed that product formation was linear with respect to time and extract concentrations for all OUC enzyme assays, and reactions were thereafter terminated at 0 and 60min, except for OCTase (0, 40min). Argininosuccinate synthase (EC 6.3.4.5) and argininosuccinate lyase (EC 4.3.2.1) were not measured because of insufficient tissue quantities. However, these enzymes are generally present in teleosts (Saha and Ratha, 1989; Felskie et al., 1998).

Carbamoyl phosphate synthetase (EC 6.3.5.5., glutamine): assays for CPSase II and III activity were conducted as described by Korte et al. (Korte et al., 1997) and Terjesen et al. (Terjesen et al., 2000). The standard reaction mixture contained, in a final volume of 300μl: 20mmoll⁻¹ ATP, 35mmoll⁻¹ MgCl₂ (found to be optimal for C. gariepinus), 21mmoll⁻¹ phosphoenolpyruvate sodium salt, 2units of pyruvate kinase, 5mmoll⁻¹ [¹⁴C]bicarbonate (4×10⁶ to 8×10⁶ctsmin⁻¹), 55mmoll⁻¹ Hepes (pH7.5), 55mmoll⁻¹ KCl, 0.5mmoll⁻¹ dithiothreitol (DTT), 0.5mmoll⁻¹ EDTA and, where indicated, 20mmoll⁻¹ glutamine, 1.7mmoll⁻¹ AGA, 133mmoll⁻¹ NH₄Cl and/or 1.7mmoll⁻¹ UTP. Assays were initiated by adding 100μl of extract.

Glutamine synthetase (GSase, E.C. 6.3.1.2) was assayed by measuring the γ-glutamyl transferase reaction (Webb and Brown, 1980) with the following modifications. Reaction mixtures contained, in a final volume of 800μl: 40mmoll⁻¹ imidazole (pH6.8), 28mmoll⁻¹ KCl, 0.6mmoll⁻¹ DTT, 0.3mmoll⁻¹ EDTA, 28mmoll⁻¹ Hepes (pH7.5), 19mmoll⁻¹ potassium arsenate, 15mmoll⁻¹ hydroxylamine, 60mmoll⁻¹ glutamine, 5mmoll⁻¹ MnCl₂ and 0.4mmoll⁻¹ ADP. Reactions were initiated with 200μl of extract.

Arginase (EC 3.5.3.1) reaction mixtures contained 57mmoll⁻¹ glycine (pH9.7), 10mmoll⁻¹ Hepes (pH7.5), 10mmoll⁻¹ KCl, 0.1mmoll⁻¹ EDTA, 0.2mmoll⁻¹ DTT, 17mmoll⁻¹ arginine and 4mmoll⁻¹ MnCl₂ in a total volume of 250μl. Reactions were initiated by adding 50μl of extract (5μl for adult liver). After terminating the reaction with 70μl of 2moll⁻¹ HClO₄ and centrifugation at 9600g for 10min, urea was determined according to the method of Rahmatullah and Boyde (Rahmatullah and Boyde, 1980).

Ornithine carbamoyl transferase (OCTase, EC 2.1.3.3) was assayed essentially as described by Xiong and Anderson (Xiong and Anderson, 1989). Final reaction mixtures contained, in a total volume of 500μl: 50mmoll⁻¹ Hepes (pH7.5), 50mmoll⁻¹ KCl, 0.5mmoll⁻¹ EDTA, 1mmoll⁻¹ DTT, 5mmoll⁻¹ carbamoyl phosphate and 10mmoll⁻¹ ornithine. Reactions were initiated by addition of 50μl of extract.

Embryonic and larval dry mass was determined, proteins precipitated and solutes extracted as described by Terjesen et al. (Terjesen et al., 1997). Samples were centrifuged at 9600g for 10min, and urea in the supernatant was assayed according to the method of Rahmatullah and Boyde (Rahmatullah and Boyde, 1980). The value of moles of urea-N is equivalent to 2×moles of urea. Uric acid was determined with uricase (Sigma 292-8) (as described by Di Stefano et al., 1992) on samples homogenized in 5% ice-cold trichloroacetic acid (TCA) (final concentration). Preliminary assays resulted in low and variable uric acid levels. To improve reproducibility, a background of 10μmoll⁻¹ uric acid was subsequently added. This modification was justified since the decline in A₂₉₃ after addition of uricase was consistently greater in samples with added uric acid than in 10μmoll⁻¹ standards. Furthermore, the added uric acid gave the same decline in A₂₉₃ whether dissolved in larval extracts or in pure 5% TCA (94% recovery), as also noted for rat liver TCA extracts (94–96% recovery; Di Stefano et al., 1992). Uric acid levels were therefore calculated by correcting for added uric acid, the change in A₂₉₃ of samples without uricase and the self-extinction of uricase.

Chemicals were of analytical grade and were purchased from Sigma Chemical Co., St Louis, MO, USA, from Merck KGa, Darmstadt, Germany, or from Boehringer-Mannheim GmbH, Mannheim, Germany. Radiochemicals were purchased from NEN Life Science Products, Belgium.

Activity is reported as μmoles of product formed or substrate consumed per minute at 26°C, except for CPSase and uricase (nmoles of product formed or substrate consumed per minute). Michaelis–Menten constants (K_m) were calculated using non-linear curve fits to the Michaelis–Menten equation. Differences between means were evaluated by analyses of variance (ANOVAs) at a significance level of P<0.05. When appropriate, Tukey’s multiple-comparison test (Zar, 1984) was subsequently employed. Results are reported as means ± s.d., except where noted.

The larvae hatched between 20 and 24h post-fertilization, and complete yolk absorption (CYA) occurred around 100h post-fertilization. Wet mass and relative water content increased up to CYA, while the fed larvae gained weight exponentially until the experiment was terminated at 314h post-fertilization (Fig.1). Air-breathing behavior commenced at approximately 200h post-fertilization. Urea-N levels increased during development up to CYA (P<0.001, N=33), but declined during starvation (Fig.2). An especially rapid increase was found close to CYA when the urea-N content doubled over a period of only 18h (Fig.2A). This increase was not due to a change in wet mass, which was stable during this period (Fig.1A). Urea-N was located mainly in the yolk at the first sampling point after hatching at a concentration of 9.1mmolkg⁻¹H₂O (calculated from data in Fig.1 and Fig.2). At the next two sampling points, the larval body contained more urea-N (51% and 62% of total, respectively) than did the yolk. Urea-N was calculated to be present in yolk at a level of 19mmolkg⁻¹H₂O at 55h post-fertilization. In fed larvae, tissue concentrations of urea-N fluctuated between 10 and 13mmolkg⁻¹H₂O, considerably higher than in adults in which, depending on the tissue, concentrations of 1.3–3.3mmolkg⁻¹H₂O were measured. Uric acid was detected at variable levels in extracts of C. gariepinus larvae. Starved larvae (100–155h post-fertilization) had a uric acid concentration of 31±16mmolkg⁻¹H₂O (N=3), while fed larvae (170–313h post-fertilization) had a concentration of 59±7mmolkg⁻¹H₂O (N=4).

Whole-animal uricase activity was low, but detectable, in embryonic samples, and activities increased up to CYA (P<0.001, N=16, Fig.3A). Starved larvae showed higher uricase activity at all sampling points compared with endogenously feeding larvae, even when adjusted for yolk mass (P<0.001, N=30, data not shown). In fed larvae, uricase activity was stable between 119 and 264h post-fertilization, but uricase activity was significantly lower at 313h post-fertilization (P<0.05), except when compared with 264h post-fertilization. Both allantoinase (Fig.3B) and allantoicase (Fig.3C) activity were detected at the embryonic stage. After hatching, allantoicase activity increased by a factor of 4 during the short period between 28 and 53h post-fertilization (Fig.3C). UGL displayed the highest in vitro activity of the uricolytic enzymes during early development (Fig.3D). Allantoinase and UGL activities increased up to metamorphosis (P<0.01, N=32), when stable levels were attained. For all uricolytic enzymes, activities were generally found to follow the same pattern with age irrespective of how the activity was calculated (i.e. whether based on wet mass, dry mass or per individual), except when calculating activities per individual fed larva (because of their large individual mass). Adult C. gariepinus tissues, especially liver, generally showed higher uricolytic activities than whole-animal larval extracts. Liver had the highest activities of all uricolytic enzymes in adult C. gariepinus, except allantoicase for which kidney and liver activity did not differ (P=0.32, N=6). Uricase activity was not observed in adult muscle, in contrast to the other uricolytic enzymes which were clearly detectable in this tissue.

Both CPSase II (pyrimidine biosynthesis) and CPSase III (OUC) were detected in C. gariepinus larvae by subjecting extracts to gel filtration chromatography (Fig.4; 313h post-fertilization). Two peaks with CPSase activity were obtained, the second peak having the higher activity, corresponding to 1.1nmolg⁻¹wetmassmin⁻¹. The second activity peak showed significant activation by AGA, low inhibition by UTP, negligible activity with NH₄Cl as a substrate and a molecular mass of 150–160kDa, all characteristics of CPSase III (Table1). In contrast, the first activity peak was presumably a CPSase II because of its lower elution volume and contrasting characteristics compared with those of the second peak (Table1). The effect of varying glutamine concentrations on the second activity peak (Fig.5) shows that this CPSase III has virtually no activity (3–6%) using standard assay conditions in the absence of its allosteric effector AGA. The apparent K_m for glutamine in the presence of 1.7mmoll⁻¹ AGA was 0.13mmoll⁻¹ but, because of the high dependence on AGA, activities were too low to estimate K_m in the absence of this effector.

The tissue distribution of CPSase III in fed C. gariepinus larvae was determined by gel-filtration chromatography (Fig.4; muscle and liver, 217h post-fertilization). The muscle elution showed two activity peaks, an almost identical elution profile to that of whole larvae at 313h post-fertilization. The second muscle activity peak eluted at 157kDa and had a total activity of 1.8nmolg⁻¹wetmassmin⁻¹, as calculated by integration. In contrast, elution of dissected liver extracts resulted in trace levels of activity. Even if these levels represent CPSase III, the total activity calculated by integration of the curve is only 0.05nmolg⁻¹wetmassmin⁻¹ or 3% of the activity in muscle.

In whole-animal measurements, CPSase II showed higher activity during the embryonic and early yolk-sac larval stages than in the later stages (Fig.6A). CPSase III was detectable in crude extracts throughout the experiment (Fig.6B), but activity was far lower than for CPSase II during the embryonic and early yolk-sac stages, as illustrated by the high degree of inhibition by UTP (Fig.6C) and lack of activation by AGA (Fig.6D) during this period. Following hatching, CPSase III activity increased to a maximum at 79h post-fertilization (Fig.6B), but declined in starved larvae (P<0.001, N=16). In fed larvae, CPSase III was the dominant form and showed highest activity at 169h post-fertilization (1.3nmolg⁻¹wetmassmin⁻¹). At 313h post-fertilization, CPSase III activity in whole-animal crude extracts was 0.9nmolg⁻¹wetmassmin⁻¹, which is comparable with that measured using gel filtration chromatography (1.1nmolg⁻¹wetmassmin⁻¹; Fig.4). In contrast, the activity in muscle calculated from activity eluted after gel filtration chromatography (Fig.4; 217h post-fertilization) was higher than in whole-animal homogenates (1.8 versus 1.0nmolg⁻¹wetmassmin⁻¹). With only minor changes, the reported CPSase activities followed the same trend with age irrespective of how they were calculated (except per individual fed larva). Assuming that yolk does not contain CPSase, fed larvae also had a higher CPSase III activity than yolk-sac larval bodies (data based on dry mass, P<0.001, N=25). Adult liver had a high CPSase II activity, but contained low and variable levels of CPSase III (Fig.6A,B). Muscle tissue did not show detectable CPSase activity. Kidney extracts of adult C. gariepinus, however, showed a slight activation by AGA and incomplete inhibition by UTP, suggesting that a low level of CPSase III activity is present in kidney tissue.

The activities of the other assayed OUC enzymes generally increased with age when results were expressed on a whole-animal mass-specific basis (Fig.7). It is noteworthy that, during the period of increased CPSase III expression in fed larvae (Fig.6B), a similar increase in OCTase activity was also observed (Fig.7B). GSase and arginase activities were higher in starved larvae than in fed larvae (P<0.01, N=18), irrespective of how activities were calculated (except μmolmin⁻¹individual⁻¹ because of the larger mass of fed larvae). The activities of the OUC enzymes varied among individuals of adult C. gariepinus (Fig.7). Pooled data did not show significant differences between tissues in OCTase and GSase expression; within the same individual, however, activities tended towards higher expression of OCTase in muscle and of GSase in liver. Arginase was clearly expressed at highest levels in liver (P<0.05, N=9).

The present study demonstrates that all three pathways for urea production, argininolysis, uricolysis and the OUC, are expressed during early development of C. gariepinus. Furthermore, the present finding (Fig.2) that urea accumulates in C. gariepinus larvae also suggests that the relatively high rates of urea excretion observed previously (Terjesen et al., 1997) represent metabolically produced urea. When the excretion data of Terjesen et al. (Terjesen et al., 1997) are employed, it can be estimated that urea accumulation accounts for less than 10% of total urea production (urea-N accumulation + excretion). Nevertheless, urea-N concentrations in fed C. gariepinus larvae are higher than in adult C. gariepinus tissues (Fig.2B), and two- to 10-fold higher than in rainbow trout and Atlantic cod larvae at first-feeding (Wright et al., 1995; Chadwick and Wright, 1999).

This is the first study to report expression of the uricolytic enzymes in early life stages of a teleost. Recently, Vigetti et al. (Vigetti et al., 2000) showed that, in Xenopus laevis embryos, allantoicase mRNA is present from 24h post-fertilization. Furthermore, xanthine dehydrogenase, the enzyme preceding uricase in purine catabolism, has recently been detected in zebrafish embryos (Danio rerio; Ziegler et al., 2000). In endogenously feeding C. gariepinus embryos and yolk-sac larvae, it would appear energetically reasonable to salvage purines originating from nucleic acid turnover instead of breaking them down through uricolysis. Possibly, in the yolk-sac larvae, the uricolytic pathway is expressed in preparation for nucleotide absorption at first-feeding, when the pathway can serve to degrade purines in excess of requirements for nucleic acid synthesis, the generally accepted role for uricolysis in adult teleosts (Wood, 1993). Measurements of purine salvage capacity in endogenously feeding stages of C. gariepinus could serve to test this hypothesis.

Uricase activity per individual is higher in C. gariepinus larvae at 123h post-fertilization (starved larvae) than in yolk-sac larvae close to CYA at both 70 and 99h post-fertilization (P<0.01, N=14). Since the larval body compartments at these sampling points were of similar size (Fig.1A), these results suggest that uricase is induced during starvation, possibly as a response to increased nucleotide breakdown. Interestingly, Rumsey et al. (Rumsey et al., 1991) found that liver uricase activity was correlated with dietary nucleic acid content in adult rainbow trout. However, the higher uricase activity during starvation of C. gariepinus larvae is not reflected in elevated urea production in vivo during this period (Terjesen et al., 1997), suggesting either that variations in uricase expression do not solely control uricolytic flux or that changes in other urea-producing pathways had masked any increased urea production through uricolysis. A similar relationship exists in adult tilapia (Oreochromis niloticus): tilapia fasted for 14 days had higher levels of liver uricase activity than fed fish even though urea excretion rates were much lower in the fasted group (Wright, 1993).

The eluted CPSase III muscle peak at 217h post-fertilization (Fig.4) showed a considerably higher total activity than whole-animal extracts at 217h post-fertilization (Fig.6), while the larval liver elution showed only trace levels of activity (Fig.4). Although protein levels were low in the larval liver elution, it can be calculated that a detectable activity peak would still elute if CPSase III were expressed to any significant extent in this tissue. Thus, CPSase III is predominantly expressed in muscle of larval C. gariepinus, as has been reported in adults of several species (Korte et al., 1997; Kong et al., 1998; Lindley et al., 1999; Terjesen et al., 2000). Larval C. gariepinus CPSase III showed considerable dependence on AGA in the reaction mixture for maximal activity, even at very high glutamine concentrations (Fig.5), and this contrasts with the situation in Atlantic halibut larvae in which AGA has less influence on activity (Terjesen et al., 2000). It appears that C. gariepinus CPSase III has nearly as high a dependence on AGA as mammalian CPSase I, for which activity in the absence of AGA is only 3% (Rubio et al., 1983). In view of this observation, short-term regulation of carbamoyl phosphate formation in C. gariepinus larvae may be influenced by changes in AGA concentrations as a result of differences in protein or amino acid absorption, as in mammals (Mejer, 1995) and as suggested for adult Gulf toadfish (Opsanus tau, Julsrud et al., 1998).

Several authors have suggested that expression of the OUC during early life stages of teleosts serves as a safeguard to control internal ammonia levels (Griffith, 1991; Wright et al., 1995; Korsgaard et al., 1995). Even though ammonia is predominantly (98%) excreted, ammonia concentrations nevertheless increase during the early yolk-sac stage in C. gariepinus, despite the small size of these larvae (Terjesen et al., 1997). Although in vivo tracer studies are required to establish whether ammonia detoxification occurs in yolk-sac larvae of C. gariepinus, the potential for urea synthesis from ammonia does exist concurrent to this ammonia accumulation since GSase, CPSase III, OCTase and arginase were expressed during this period (Fig.6, Fig.7). Adult C. gariepinus, however, had only low and variable levels of CPSase III activity (Fig.6), suggesting that the expression of OUC enzymes is influenced by developmental stage and, further, that any urea produced by adult C. gariepinus under the experimental conditions employed here is produced predominantly through uricolysis and/or argininolysis. In contrast, adults of its Asian relative C. batrachus had high activities of the OUC enzymes even in the absence of environmental challenges (Saha and Ratha, 1989; Saha et al., 1999). Still, C. gariepinus adults must also have a mechanism to detoxify or eliminate ammonia efficiently, given their high LC₅₀ at 96h of 380μmoll⁻¹ NH₃ (Britz, 1988; Oellermann, 1995) and their ability to survive for 2 months in mud (Donnely, 1973).

The present study has shown that the enzymes necessary for urea production through argininolysis (arginase), through the uricolytic pathway or through the OUC are expressed during the early life stages of C. gariepinus. Field studies of variables related to ammonia excretion, such as pH and ammonia levels, could further reveal whether OUC expression during the early life stages of C. gariepinus constitutes an adaptive mechanism to the fluctuating availability of water at the spawning sites (Greenwood, 1955).

B.F.T. is grateful to Dr P. M. Anderson and two anonymous referees for helpful comments on this manuscript, and to Dr Armando Garcìa-Ortega and the staff at WIAS for excellent assistance with catfish rearing. This study was funded by the Grant for Biological Research provided by the University of Bergen to B.F.T., by a Premier’s Research Excellence Award to P.A.W. and by the Research Council of Norway (project no. 115876/122) to I.R.