Ammonia detoxification and localization of urea cycle enzyme activity in embryos of the rainbow trout (Oncorhynchus mykiss) in relation to early tolerance to high environmental ammonia levels

Recently, significant activities of the ornithine–urea cycle (OUC) enzymes carbamoyl phosphate synthetase III (CPSase III), ornithine transcarbamoylase (OTCase) and arginase, and of the accessory enzyme glutamine synthetase (GSase), have been measured in the embryos of the freshwater rainbow trout (Oncorhynchus mykiss) and the marine Atlantic cod (Gadus morhua) and halibut (Hippoglossus hippoglossus), whereas activities are either low or absent in adults (Wright et al., 1995; Chadwick and Wright, 1999; Terjesen et al., 2000). The localization of these enzymes within the developing embryos has not been determined. Urea synthesis and the OUC typically occur in the liver of ureotelic vertebrates (for reviews, see Campbell, 1991; Anderson, 1995). Recently, significant levels of the first two enzymes in the OUC, CPSase III and OTCase, have been detected in adult skeletal muscle tissue of rainbow trout (Korte et al., 1997), gulf toadfish (Opsanus beta) (Julsrud et al., 1998), largemouth bass (Micropterus salmoides) (Kong et al., 1998) and most notably the alkaline-lake-adapted tilapia (Oreochromis alcalicus grahami) (Lindley et al., 1999). In rainbow trout, CPSase III is completely absent from adult liver tissue (Korte et al., 1997). Does a similar pattern of OUC expression occur in embryonic trout? In previous studies on whole rainbow trout embryos, OUC enzyme activities were not detected prior to hatching, possibly because intact embryos consist mostly of yolk with a much smaller percentage of embryonic tissue (Wright et al., 1995). We hypothesized that the yolk is devoid of OUC enzyme activities and therefore that previous analysis of enzyme activities in whole embryos underestimated the absolute values of enzyme activities in the embryonic tissues. Moreover, by determining the tissue distribution of the urea cycle enzymes in embryos, we could then optimize the protocol for analyzing enzyme activities in relatively small tissue specimens.

In post-yolk-sac embryos of the viviparous blenny Zoarces viviparus, urea is the predominant nitrogenous waste product (Korsgaard, 1994; Korsgaard, 1997). Urea excretion can account for a significant amount of nitrogen excretion in oviparous teleost embryos, but ammonia is the primary nitrogenous waste product (e.g. Torres et al., 1996; Chadwick and Wright, 1999; Pilley and Wright, 2000). Diffusion of ammonia out of the encapsulated embryo may be hindered by the absence of respiratory convection (i.e. gill ventilation) and indirect contact with bulk water. Indeed, ammonia levels rise steadily in both freshwater (Wright et al., 1995; Terjesen et al., 1997) and marine (Fyhn and Serigstad, 1987; Finn et al., 1996; Chadwick and Wright, 1999) species during embryogenesis.

Are elevated tissue levels of the potentially toxic ammonia one of the triggers for hatching? Environmental factors (e.g. hypoxia, temperature, light, pH) and developmental stage are important in the initiation of hatching in fishes (DiMichele and Taylor, 1980; Yamagami, 1988). Hatching may facilitate the removal of ammonia from the tissues by enabling the embryos to have direct contact with the external environment, which functions as an unlimited sink for the diffusive loss of ammonia. Thus, a second aim of this study was to examine the effects of elevated external ammonia levels on the time to hatching.

Considering the relative toxicity of ammonia, it is surprising that, compared with their adult forms, the embryos of many teleost species are highly tolerant to external ammonia. Rice and Stokes (Rice and Stokes, 1975) reported that rainbow trout embryos had a 24h tolerance limit of greater than 3.58mgl⁻¹ of un-ionized ammonia [approximately 2.25mmoll⁻¹ (NH₄)₂SO₄] compared with the value of 0.097mgl⁻¹ [approximately 0.06mmoll⁻¹ (NH₄)₂SO₄] in adult fish. Similar trends have been observed in the green sunfish (Lepomis cyanellus) (McCormick et al., 1984), the spotted seatrout Cynoscion nebulosus (Daniels et al., 1987) and the smallmouth bass Macropterus dolomieni Lacepede (Broderius et al., 1985). It is possible that this difference in ammonia tolerance between early and late stages is related to characteristics that are unique to the embryo. The protection afforded by the chorion may not be a factor because just-hatched embryos are less susceptible to ammonia than are encapsulated embryos (Rice and Stokes, 1975; Broderius et al., 1985). In fact, spotted seatrout exhibit an increase in ammonia tolerance over the 4 month period following hatching (Daniels et al., 1987). Embryos may possess efficient biochemical mechanisms by which they convert ammonia to non-toxic substances, such as the non-essential amino acids glutamine and glutamate. Adult rainbow trout (Vedel et al., 1998), mudskippers (Periophthalmus cantonens) (Iwata, 1988) and goldfish (Carassius auratus) (Levi et al., 1974) increase glutamine/glutamate production upon exposure to elevated external ammonia levels, particularly in the brain. There is little or no information available on changes in amino acid levels in relation to external ammonia exposure in fish embryos, and therefore this was an area of particular interest in the present study.

Griffith (Griffith, 1991) proposed that the OUC would have been retained and expressed at the embryonic stage to deal with the naturally increasing tissue ammonia levels incurred during embryogenesis. If this is true, then embryos may increase OUC enzyme activity as a detoxifying step in the event of elevated ammonia exposure.

The objectives of this study were threefold: (i) to identify the tissue distribution of CPSase III and other OUC enzymes within the embryo, (ii) to determine whether exposure to high external ammonia levels initiates hatching, and (iii) to determine whether embryos detoxify ammonia to other nitrogenous compounds (e.g. urea, amino acids). To address the first objective, just-hatched trout embryos were dissected by hand to separate yolk, liver and the embryo proper. The activities of CPSase III, OTCase, arginase and GSase were measured in each of these tissue fractions. For the second and third objectives, two levels of ammonia exposure were used, an acute exposure (2h) to 10mmoll⁻¹ NH₄Cl and a chronic exposure (4 days) to 0.2mmoll⁻¹ NH₄Cl. The time to hatching, ammonia and urea excretion rates and tissue levels of ammonia and urea were measured in both groups, and tissue levels of amino acids were measured in the chronic exposure group. In these analyses, it was imperative that the relatively large yolk mass was separated from the smaller embryonic tissue mass, particularly for the amino acid measurements, because yolk levels of amino acids are approximately seven times those in the embryonic tissues. Also, to delineate the role of the OUC in ammonia detoxification, the activities of CPSase III, OTCase, arginase and GSase in embryonic tissues were compared between the chronically exposed and control embryos.

Rainbow trout Oncorhynchus mykiss (Walbaum) embryos were purchased from Rainbow Springs Trout Farm (Thamesford, Ontario, Canada) and held in continuous-flow incubation trays with mesh bottoms supplied with local well water (10–12°C; pH7.9; water hardness 411mgl⁻¹ as CaCO₃; Ca²⁺, 5.24mequivl⁻¹; Cl⁻¹, 1.47mequivl⁻¹; Mg²⁺, 2.98mequivl⁻¹; K⁺, 0.06mequivl⁻¹; Na⁺, 1.05mequivl⁻¹) at Hagen Aqualab, University of Guelph, Guelph, Ontario, Canada. Incubation trays were shielded from the light during the entire incubation period. For series I experiments, hatching occurred between 32 and 34 days post-fertilization. For series II and III, experimentation was initiated at the ‘eyed-up’ stage (13 days post-fertilization in series II, 30 days post-fertilization in series III), when the pigmented eye was first clearly visible. Embryos for each series were obtained from four or more females. For the three series of experiments, separate batches of embryos were used. Hence, comparisons were not made between series, only within series. Mortality was 15% over the experimental period, and there was no significant difference in mortality between control and experimental animals.

Just-hatched embryos were anaesthetized in 0.15gl⁻¹ MS-222 for 10min, rinsed with water and placed under a dissecting microscope at 12.5× magnification. The yolk sac was teased away from the embryonic body and intact liver, and the outer yolk membrane was cut. The intact yolk sac was then removed with forceps, placed in an aluminium container and frozen in liquid nitrogen. Both the yolk and the yolk sac membrane were included as the yolk sac fraction. Next, the liver was separated from the embryonic body by cutting the primary hepatic artery and the other connecting tissues. Visual observation indicated minimal loss of tissue during this process and no cross-contamination of samples taken from individual fish. The embryonic body and isolated liver were frozen as separate fractions in liquid nitrogen. Five pooled samples of each tissue were collected, and each sample contained the tissues from 30–50 individuals. Samples were weighed and then stored at −80°C. Controls were included to confirm that MS-222 treatment did not affect enzyme activities.

Thirty-two embryos were placed into perforated polystyrene chambers. Each chamber (approximately 50ml) represented a pooled sample (N=1). In preliminary experiments, several concentrations of NH₄Cl were tested to determine what external ammonia concentration was required to induce a significant elevation of internal ammonia levels. The level of 10mmoll⁻¹ NH₄Cl was chosen because whole-embryo ammonia levels were significantly elevated (control 1.83±0.80μmolNg⁻¹versus treated 3.87±0.13μmolNg⁻¹; P=9.5×10⁻⁸, single factor ANOVA) after 2h of exposure. Six treated chambers (10mmoll⁻¹ NH₄Cl) and six control chambers (fresh water) were assigned to each time period, 0, 1, 6, 24, 48 and 120h post-treatment. Water in the experimental chambers was aerated throughout the experiment.

Embryos were exposed to either treatment or control water for 2h and then returned to fresh water. The pH of the NH₄Cl solution was adjusted to match that of the control solution (pH7.94). Immediately following treatment (0h), a sub-sample of 10 embryos from each of six treated and six control chambers was collected, blotted dry, weighed and immediately frozen in liquid nitrogen for later tissue analysis of urea and ammonia levels. Ammonia and urea excretion rates over 2h were measured at the end of each time period (as described previously, Wright et al., 1995).

In eight separate chambers (four treated, four control) containing 25 embryos each, post-treatment or control animals were monitored for time to hatching. Hatching was recorded on a daily basis until it was complete.

In the chronic ammonia exposure experiment, it was necessary to choose a concentration of external NH₄Cl well below the level in the acute experiment, but sufficient to cause a small but marked disturbance in internal ammonia levels. We chose an external level of 0.2mmoll⁻¹ NH₄Cl on the basis of previous experiments (Wright and Land, 1998). Forty embryos were placed in perforated polystyrene chambers. Each chamber held a volume of approximately 250ml and represented one pooled sample (N=1). Twelve chambers were allocated to either the control (fresh water, N=6) or treated (0.2mmoll⁻¹ NH₄Cl, N=6) groups. As before, chambers were continuously aerated throughout the 4 day treatment period. The pH of the NH₄Cl solution was adjusted to match that of the control solution (pH7.86). Solutions were changed in both groups every 12h to avoid accumulation of endogenously excreted ammonia (levels remained at <20μmoll⁻¹).

Nitrogen excretion rates over 3h were measured on day 0 (at the onset of treatment) and on days 1, 2 and 4 of treatment in both the control and treated embryos, as described for series II. Tissue samples were taken at the end of day 4 of the treatment period. Embryos were removed from the chambers, rinsed briefly in fresh water, blotted dry and frozen in liquid nitrogen for later tissue analysis of ammonia, urea and amino acid levels and urea cycle enzyme activities.

Hatching was monitored as described for series II.

Water samples were stored at −20°C for up to 4 weeks prior to analysis. Water samples from both series II and III were analyzed for ammonia content using colorimetric assays (as described by Verdouw et al., 1978). The detection limits of this assay were 0.1–100μmoll⁻¹ ammonia. For series III, water samples from treated chambers (0.2mmoll⁻¹ NH₄Cl) were diluted 2.5-fold before being assayed for ammonia concentration. These samples were also assayed in triplicate to reduce error associated with a high background level of ammonia. Urea content in series III was measured using colorimetric assays (as described by Rahmatullah and Boyde, 1980). The detection limits of this assay are 0.2–200μmoll⁻¹ urea. All spectrophotometric measurements in this study were performed using a Perkin Elmer UV/VIS spectrophotometer (Lambda 2) (Perkin Elmer Corp., Norwalk, CT, USA). Excretion rates were calculated as the difference in concentration between the initial and final water sample multiplied by the volume of the chamber and divided by the total wet mass of the animals in the chamber and the excretion time.

To separate the embryonic tissue (including the yolk sac membrane) from the yolk for further analyses of ammonia, urea and amino acid levels and enzyme activities, a centrifugation method was used (A. Shashsavarani, Z. Thomas, J. S. Ballantyne and P. A. Wright, in preparation). The chorions of five embryos per sample were first removed manually using fine forceps. The embryos were diluted in five times their mass:volume of 50mmoll⁻¹ imidazole/50mmoll⁻¹ potassium chloride buffer (pH8, 4°C). Samples were gently inverted until all the yolk was completely dissolved in the buffer and the remaining embryos were floating freely in the solution and were centrifuged briefly (3s). The supernatant was removed and stored as the yolk fraction at −80°C. Embryos were rinsed again in imidazole/KCl buffer and frozen at −80°C.

Yolk and embryo fractions were stored at −80°C for up to a week prior to analysis. The embryo fractions were ground using a mortar and pestle under liquid nitrogen, deproteinized and neutralized as described by Wright et al. (Wright et al., 1995) with the following exceptions; 10 volumes of ice-cold 8% perchloric acid (PCA) was used to deproteinize each sample, followed by centrifugation for 10min at 16000g. Yolk fractions (250μl of the total fraction) were deproteinized using twice the volume of ice-cold 8% PCA and centrifuged for 10min at 16000g. Ammonia content was measured in all fractions using an enzymatic Sigma diagnostic kit (171-C, Sigma-Aldrich Canada Ltd, Oakville, Ontario, Canada), and urea concentration was analyzed using the method of Rahmatullah and Boyde (Rahmatullah and Boyde, 1980). The detection limits of the Sigma diagnostic kit are 1–882μmoll⁻¹ ammonia.

Separated yolks and embryos (series III) were stored at −80°C for up to 2 weeks prior to amino acid analysis. Analysis was performed using high-performance liquid chromatography (HPLC). Yolk samples (a 200μl fraction of each pooled yolk sample) were analyzed using the solvents and method described by Barton et al. (Barton et al., 1995) for plasma. Embryo samples were analyzed using the same solvents, but samples were processed by grinding approximately 125mg of tissue in 500μl of 0.5% trifluoroacetic acid in methanol. Samples were brought up to 750μl using double-distilled water. A 6mmoll⁻¹ internal standard of l-norvaline and l-azetidine-2-carboxylic acid was used for the embryo samples (12.5μl per sample). Samples were then centrifuged for 5min at 4°C at 16500g, after which 50μl of 1moll⁻¹ sodium acetate and 25μl of 1moll⁻¹ sodium hydroxide was added to each sample. Samples were then centrifuged for an additional 25min at 4°C and 16500g. The resulting supernatant was analyzed using a Hewlett Packard Series II 1090 liquid chromatograph (Hewlett-Packard Company, Avondale, PA, USA).

For urea cycle enzyme analysis (series III), embryos were separated as described above except that 14 embryos per sample (0.15–0.25g) were used and the yolks were discarded. Embryos used for enzyme analysis were stored overnight at −80°C prior to analysis. Samples were homogenized in approximately 20 volumes of ice-cold extract buffer (0.05moll⁻¹ Hepes buffer, pH7.5, 0.05moll⁻¹ KCl, 0.5mmoll⁻¹ EDTA, 1mmoll⁻¹dl-dithiothreitol) and sonicated. Homogenates were centrifuged at 4°C for 10min at 14000g. Because of the relatively low mass of samples, the supernatant from the centrifuged tissue homogenate was not passed through a Sephadex column as described previously (Felskie et al., 1998). To account for the presence of endogenous substrates in the homogenate, controls without exogenous substrate were included in each enzyme assay.

All enzyme activities were measured at 26°C. OTCase activity was determined by measuring the production of citrulline using the reaction mixture described by Wright et al. (Wright et al., 1995) and the colorimetric assay method described by Xiong and Anderson (Xiong and Anderson, 1989) from 0 to 40min. GSase activity was quantified as the production of γ-glutamyl hydroxamate from 0 to 40min using the method described by Shankar and Anderson (Shankar and Anderson, 1985). Arginase activity was measured using the reaction mixture described by Felskie et al. (Felskie et al., 1998) and the colorimetric assay described by Rahmatullah and Boyde (Rahmatullah and Boyde, 1980). The reaction mixture used for CPSase activity measurement was as described in Chadwick and Wright (Chadwick and Wright, 1999), and the reaction was conducted for 60min, after which [¹⁴C]carbamoyl phosphate production was measured using the technique described by Anderson et al. (Anderson et al., 1970).

For series I, an additional set of reactions was included in which NH₄Cl was substituted for glutamine (at the same concentration) to compare the efficacies with which CPSase III uses ammonia or glutamine as substrates. Although it is generally accepted that ammonia is not a physiologically significant nitrogen-donating substrate for the urea-cycle-related CPSase in fish (Felskie et al., 1998), recent work on fish tissues has shown that ammonia can be as important as or more important than glutamine (Saha et al., 1997; Lindley et al., 1999). Activities were expressed as the number of μmoles of product formed per gram of tissue per minute (μmolg⁻¹min⁻¹). CPSase III activity was defined as the activity in the presence of glutamine (the substrate), N-acetylglutamate (AGA, a required positive effector) and uridine-3′-triphosphate (UTP, a CPSase II inhibitor). The estimated limit of detection for this assay is 0.1nmolg⁻¹min⁻¹. CPSase II was defined as activity in the presence of glutamine and the absence of AGA.

The differences between control and ammonia-treated values for the yolk and embryonic tissue analysis (ammonia, urea and enzyme activities) and time to hatching within each series were compared using single-factor analysis of variance (ANOVA). Significant differences were declared if P<0.05. Excretion data were analyzed using repeated-measures ANOVA and amino acid data using a General Linear Models procedure, both using the SAS system (version 6.12; SAS Institute Inc., Cary, NC, USA). The Tukey test was applied after significant differences had been identified between treatment and control animals (P<0.05). Results are presented as means ± s.e.m.

No CPSase activity was detected in the yolk sac of rainbow trout (Fig.1). The absence of CPSase II was indicated by the lack of activity in the presence of substrate, and the absence of CPSase III was indicated by the lack of activation by AGA. In the embryonic body, there was substantial CPSase III activity, as indicated by the activation of activity (106%) in the presence of AGA (glutamine alone, 0.79±0.12nmolg⁻¹min⁻¹versus glutamine+AGA, 1.63±0.14nmolg⁻¹min⁻¹, N=5). CPSase II activity, but not CPSase III activity, was detected in the liver. In the embryonic body and liver, the activity of CPSase II with ammonia as a substrate was as high as that with glutamine. The absence of activation by AGA (ammonia alone, 0.69±0.09nmolg⁻¹min⁻¹versus ammonia+AGA, 0.78±0.08nmolg⁻¹min⁻¹, N=5) indicated that ammonia was not used as a substrate by CPSase III. With respect to the whole animal, CPSase III activity in the embryonic body (minus the liver) accounted for 100% of the total CPSase III activity. CPSase II activity was almost three times greater in the liver than in the embryonic body (Fig.1).

GSase, OTCase and arginase activity was detected in all fractions, but the activities of these enzymes were relatively low in the yolk sac (Fig.1). OTCase activity was four times greater in the liver than in the embryonic body (Fig.1).

Hatching in series II began 23 days post-fertilization and took 6 days to complete, whereas in series III, hatching began 41 days post-fertilization and took 7 days to complete. Time to hatching was not affected by treatment with NH₄Cl in either series (Fig.2).

In the series II experiments, ammonia excretion was significantly higher (+62%) in ammonia-treated embryos (1h) than in control animals immediately after the treatment (time 0h). Ammonia excretion rates remained significantly higher in treated embryos up to 24h post-treatment, but declined to match those of controls by 48h (Fig.3A). Urea excretion rates were significantly elevated immediately following treatment (0h), but returned to control values by 24h (Fig. 3B).

Ammonia excretion in series III was measured during treatment (0.2mmoll⁻¹ NH₄Cl) as opposed to after treatment as in series II. Ammonia excretion was significantly depressed on the first day of treatment (day 0) and was, in fact, negative, indicating an uptake of ammonia into the embryos (Fig.4A). Ammonia excretion recovered by day 2. Urea excretion in ammonia-exposed embryos (series III) was significantly higher than in controls only on day 4 (Fig.4B).

In series II and III, ammonia levels in the yolk were significantly higher (+102% and +304%, respectively) than control yolk levels following either acute or chronic ammonia exposure (Fig.5A). In the embryonic tissues, however, the ammonia concentration was not altered significantly by these treatments. Urea levels in the yolk were unchanged by acute or chronic ammonia exposure, but they were significantly elevated in the embryonic tissues (Fig.5B).

There were distinct differences in ammonia and urea levels in different tissue compartments. Particularly in ammonia-exposed embryos, ammonia concentration in the yolk fraction was several-fold higher than in the embryo fraction (Fig.5A). In all groups, urea levels in the yolk fraction were three- to fivefold higher than in the embryo fraction (Fig.5B).

There was no effect of chronic exposure (4 days) to 0.2mmoll⁻¹ NH₄Cl on urea cycle enzyme activities in series III embryos (Table1).

Total amino acid levels in the embryonic tissues were not significantly different between control and ammonia-exposed (chronic) embryos (Table2). Levels of sixteen individual essential and non-essential amino acids in the yolk, however, were significantly lower (−34% to −58%) after 4 days of exposure to 0.2mmoll⁻¹ NH₄Cl. There were no differences between individual amino acids in embryonic tissues with ammonia exposure.

This study provides evidence of urea cycle expression in non-hepatic tissues of rainbow trout just after hatching because CPSase III, GSase, OTCase and arginase are co-localized to the embryonic body (yolk sac and liver removed). These findings are consistent with previous reports of CPSase III expression in skeletal muscle tissue of juvenile and adult rainbow trout, but not in liver tissue (Felskie, 1996; Korte et al., 1997). In addition, the data support our hypothesis that enzyme analysis on whole-animal homogenates will underestimate enzyme activity in the embryonic tissues because of the presence of the relatively large yolk containing low or undetectable activities of all four measured enzymes. The enzyme activities measured in the embryonic body and liver in the present study are consistently higher than in the whole-animal homogenates of approximately the same developmental stage (just hatched) used by Wright et al. (Wright et al., 1995). Indeed, OTCase activity is 200-fold higher in the liver (present study) than in the whole animal (Wright et al., 1995). It is also interesting that the level of CPSase III activity in the embryonic body (consisting largely of skeletal muscle) is relatively high (two- to sevenfold higher) compared with juvenile (Felskie, 1996) and adult (Korte et al., 1997) trout muscle tissue.

The presence of relatively high levels of OTCase activity in liver tissue of newly hatched trout is difficult to reconcile with the absence of CPSase III activity (CPSase III provides the substrate for OTCase), indicating the lack of a functional hepatic urea cycle in hatched trout. OTCase is not thought to be involved in other metabolic reactions. Felskie et al. (Felskie et al., 1998) suggested that urea cycle intermediates may be shuttled between tissues (e.g. from muscle to liver) in some fishes, but this has not been demonstrated. In the alkaline-lake-adapted tilapia Oreochromis alcalicus grahami, the activities of all the urea cycle enzymes are present in the skeletal muscle tissue at levels that surpass those measured in the liver (Lindley et al., 1999). At least in the ureotelic tilapia, urea synthesis probably occurs primarily in the muscle tissue. Further studies in trout embryos using a radiolabelled substrate for the urea cycle would be valuable in sorting out which tissues are involved in urea-cycle-related urea synthesis.

Hatching is not affected by treatment with either 0.2mmoll⁻¹ (4 day) or 10mmoll⁻¹ (2h) NH₄Cl, consistent with a study on Atlantic salmon (Oncorhynchus gorbuschka) in which embryos were exposed to 7–11mmoll⁻¹ (NH₄)₂SO₄ (24h) (Rice and Bailey, 1980). It does appear, therefore, that external ammonia, even at quite high levels, does not directly initiate the hatching process. In our preliminary analysis of the ammonia content of intact (whole) embryos, there was a significant elevation in ammonia levels after 2h of exposure to 10mmoll⁻¹ NH₄Cl (control 1.83±0.08μmolg⁻¹, N=6, versus treated 3.87±0.13μmolg⁻¹, N=6). Subsequent separation of the yolk and tissue fractions, however, reveals that neither acute nor chronic ammonia treatment results in a significant elevation of ammonia concentration in the embryonic tissues. It is still possible that sustained elevation of internal ammonia levels during embryogenesis may influence the time to hatching in trout. Moreover, although environmental factors influence hatching (see Introduction), the exact developmental stage of exposure may be critical and is a consideration for future study.

The present data demonstrate that trout embryos have a remarkable ability to tolerate environmental ammonia. This tolerance is not related to impermeability because ammonia is clearly absorbed by the embryos from the environment. For example, the significant elevation in the rate of ammonia excretion for 24h following the acute 10mmoll⁻¹ NH₄Cl treatment indicates that excess ammonia is cleared from the embryos once they have been placed in ammonia-free water. During chronic exposure to 0.2mmoll⁻¹ NH₄Cl, ammonia excretion rates are initially reversed (day 1), indicating uptake of ammonia from the external water to the animal. Although ammonia levels in the embryonic tissues are unaltered in both groups, the yolk sac is clearly a site of ammonia storage. Yolk ammonia levels increase two- to threefold in acute and chronic exposures, respectively. Terjesen et al. (Terjesen et al., 1998) reported a twofold increase in the ammonia content of intact embryos of the marine Atlantic halibut Hippoglossus hippoglossus after 6 days of exposure to exceedingly high ammonia levels (27mmoll⁻¹ NH₄Cl). Both studies highlight the extreme tolerance to environmental ammonia demonstrated by some species at the embryonic stage.

We propose that ammonia tolerance is related, in part, to mechanisms that maintain low and relatively constant ammonia concentrations in the embryonic tissues. These mechanisms may involve detoxification of ammonia by synthesis of urea and sequestration of ammonia away from the developing tissues, i.e. in the yolk sac. Urea levels in the tissues are significantly elevated in both acute and chronic NH₄Cl treatment groups. In the chronically exposed embryos, urea excretion rates are also significantly higher than in the control group. The elevation of urea concentrations in the tissues, but not the yolk, is consistent with the expression of CPSase III, OTCase, arginase and GSase in the embryonic tissues. The levels of enzyme activity are not altered by chronic ammonia exposure. The maximal activity of the rate-limiting enzyme in a pathway can be compared with the rate of product formation to determine whether the level of activity is within the expected physiological range. In this case, CPSase III activities per individual are compared with the rate of urea excretion per individual (series III), presumably a valid comparison because very little excess urea is retained in the yolk or tissue fraction (approximately 5%) during chronic ammonia exposure and most appeared to be excreted. With the above assumptions, CPSase III activity can account for approximately 70% of the urea excreted under control conditions, but only 52% during NH₄Cl exposure. Thus, approximately half the urea synthesized under hyperammonia stress may be derived from alternative sources such as de novo purine synthesis and subsequent uric acid degradation or arginolysis (e.g. Wright, 1993; Wright and Land, 1998).

Another possible pathway for temporary storage of ammonia is through the synthesis of the non-essential amino acids glutamine and glutamate. In embryos chronically exposed to ammonia (4 days), there are no changes in tissue glutamine or glutamate levels or in total amino acid levels, but there are significant decreases in the yolk concentrations of almost every amino acid measured. These findings are puzzling. Dilution of the yolk compartment could potentially explain these findings, but yolk water content was not significantly different between these two groups (61–63% water by mass). It is unlikely that amino acids were lost to the environment because the yolk sac membrane is rather impermeable to solutes (e.g. Potts and Rudy, 1969; Mangor-Jensen et al., 1993), and developmental decreases in yolk amino acid concentrations are not due to diffusive loss (Rønnestad, 1993). Likewise, amino acids are not transferred for storage in the embryonic tissues (Table2). Thus, the decrease in amino acid levels in ammonia-exposed embryos may reflect catabolism (see below) or, less importantly, incorporation into yolk proteins. We propose the latter possibility to explain the disappearance of taurine, a non-metabolisable amino acid.

To understand more thoroughly the changes in nitrogen equivalents during chronic ammonia exposure, a nitrogen budget was calculated. Over the 4 days of exposure, an individual embryo absorbed 0.31μmol-N of ammonia from the environment, and 0.05μmol-N or approximately 16% of this amount was converted to urea. The difference in the yolk ammonia content between control and ammonia-exposed embryos is 0.43μmol-N; this level is somewhat higher than can be accounted for simply by absorption from the environment. The difference may be due, in part, to catabolism of individual amino acids. Theoretically, approximately 0.9μmol-N of ammonia would be produced if the significant decreases in yolk amino acid levels observed (Table2) were as a result of complete amino acid catabolism. The difficulty with this hypothesis is that amino acid catabolism would occur mostly in the tissues, but no excess tissue ammonia was detected, nor was the excess ammonia excreted. Thus, we propose that elevated ammonia levels in the yolk are derived primarily from the environment, with possibly a smaller proportion derived from amino acid catabolism. It should be noted, however, that stress-induced release of cortisol in adult teleosts alters protein and amino acid metabolism as well as increasing the rate of ammonia excretion to the environment (Mommsen et al., 1999). Whether cortisol is involved in hyperammonia stress in trout embryos is unknown.

Rahaman-Noronha et al. (Rahaman-Noronha et al., 1996) examined ammonia distribution in the yolk versus the perivitelline fluid in rainbow trout on the basis of differences in pH and electrical potential and found that compartments with a lower pH (e.g., yolk) support higher concentrations of ammonia. Likewise, the low pH of the uterine fluid surrounding the pup of the dogfish (Squalus acanthias) provides a sink for ammonia eliminated from both the mother and the pup (Kormanik and Evans, 1986). To predict yolk total ammonia concentration ([T_amm]) and determine whether the yolk ammonia concentration is dependent on the H⁺ gradient, we used a modification of the Henderson–Hasselbalch equation (Rahaman-Noronha et al., 1996), assuming that yolk pH is 6.35 (Rahaman-Noronha et al., 1996), that extracellular pH (pHe) is 7.8 and that intracellular[T_amm]:extracellular[T_amm] is 22 (Wilkie and Wood, 1991). The predicted control yolk [T_amm] is 2.71μmol-Ng⁻¹, not very different from the measured value of 2.56μmol-Ng⁻¹ (Fig.5A), suggesting that the yolk:tissue ammonia distribution was probably dependent on the H⁺ gradient. When these calculations were repeated using values for NH₄Cl-exposed embryos, however, the predicted yolk T_amm was three- to fourfold lower than the measured values. In other words, although yolk ammonia levels are significantly elevated in NH₄Cl-treated embryos, the absolute level is not as high as expected on the basis of the measured tissue ammonia level and estimated H⁺ gradient. This implies that, in the presence of external ammonia, the above assumptions are either no longer valid (e.g. the NH₄Cl treatment resulted in changes in the pHe and/or yolk pH) or the yolk [T_amm] is influenced by other factors, such as the electrical potential (Rahaman-Noronha et al., 1996). It is reasonable to propose that an alkalization of the embryo and/or yolk as a result of NH₃ entry (and conversion to NH₄⁺) would have altered the H⁺ gradient between the yolk and embryo and, consequently, the yolk total ammonia concentration.

The urea content of the yolk is significantly higher than that of the embryonic tissues, with the ratio ranging from 3:1 to 6:1 (yolk:embryo; Fig.5B). Urea, unlike NH₄⁺, is not an electrolyte and its concentration should therefore be relatively uniform and not influenced by pH or electrochemical gradients. In fact, Raymond and DeVries (Raymond and DeVries, 1998) measured a uniform distribution of urea between muscle, serum, urine and liver in a variety of arctic teleosts. Rahaman-Noronha (Rahaman-Noronha, 1996) measured yolk urea concentration in trout embryos at the same stage of development, after manually removing yolk samples, and obtained very similar values to those in the present study. The large diffusion gradient for urea between the yolk and tissues suggests that urea transport solely by simple diffusion is unlikely. Pilley and Wright (Pilley and Wright, 2000) reported that urea excretion in rainbow trout embryos was dependent, in part, on a facilitated urea transporter, as in other fish tissues (Wood et al., 1998; Smith and Wright, 1999; Fines et al., 2001, Walsh et al., 2000). Clearly, research focused on the distribution of urea in the tissues of fish embryos is warranted.

In summary, CPSase III and other related OUC enzymes in rainbow trout embryos are expressed together in the embryonic body, with OTCase, arginase and GSase (but not CPSase III) detected in the liver tissue and, to a small extent, in the yolk. During acute or chronic hyperammonia stress, tissue ammonia levels remain constant and the time to hatching is not altered, although a considerable ammonia load is absorbed from the environment. Most of this excess ammonia is found in the yolk, whereas a small percentage is converted to the relatively non-toxic nitrogen end-product urea. These strategies may contribute to the high tolerance of early life stages compared with adult trout to excessive environmental ammonia levels. Furthermore, the data strongly indicate that analyses of enzyme activities and metabolite levels in fish embryos with a relatively large yolk mass (e.g. salmonids) requires the careful separation of the yolk from the embryonic tissues.

Funding for this project was provided by an NSERC Research Grant to P.A.W. and an NSERC Summer Student Research Assistantship to S.L.S.