Homologous Recombinant Growth Hormone and Calcium Metabolism in the Tilapia, Oreochromis Mossambicus, Adapted to Fresh Water

A characteristic of the members of the prolactin gene family is their pleiotropic nature. In fish, growth hormone (GH) acts as a growth-promoting hormone, but it also appears to be involved in osmoregulation, stress adaptation and reproduction (Le Bail et al. 1991). Growth, osmoregulation, stress adaptation and reproduction are strongly interdependent in fish, and all these processes depend on a well-regulated calcium balance; Ca²⁺ uptake via the gills is pivotal in calcium handling by fish (Flik et al. 1985). The interdependence of these processes and their dependence on GH may underlie the pleiotropic character of the effects of GH in fish.

Studies on the effects of GH in fish have focused on salmonids, eels, cyprinids and tilapia. The emphasis of the work carried out so far has been on the establishment of sensitive assays for the determination of (plasma) levels of the hormone (e.g. Fryer, 1979; Cook et al. 1983; Bolton et al. 1987; Le Bail et al. 1991; Ayson et al. 1993), on the demonstration of receptors for GH (Hirano, 1991; Gray et al. 1990; Sakamoto and Hirano, 1991) and on the regulation of GH-producing cells (Marchant et al. 1987; Nishioka et al. 1988). A pivotal target for GH is the liver (Hirano, 1991), where it stimulates receptors with nanomolar affinity. Liver cells activated by GH produce a fish insulin-like growth factor (IGF, probably IGF-I; Drakenberg et al. 1989) that may subsequently stimulate growth (Duan and Hirano, 1990; Sakamoto and Hirano, 1991; Gray and Kelley, 1991; McCormick et al. 1992a,b). IGF also has osmoregulatory actions, as has recently been detected in rainbow trout by McCormick et al. (1991). The demonstration of GH receptors in rainbow trout gills (Sakamoto and Hirano, 1991) and of an osmoregulatory role for IGF suggests that an analogous ‘GH–IGF system’ is present in gills and liver. The apparent low density of GH receptors in branchial epithelium (3.1fmolmg⁻¹ protein) should be interpreted with care: assuming that the ionocytes in the gills are the target for GH and an average ionocyte density of 3% in the branchial epithelium (G. Flik, personal observation on freshwater rainbow trout and tilapia), the receptor density of ionocytes (3.1X100/3=103fmolmg⁻¹ protein) may be even higher than that of hepatocytes (71.0fmolmg⁻¹ protein). Clearly, the presence of GH receptors on ionocytes is in good agreement with the consensus that GH exerts osmoregulatory effects in seawater-adapting and seawater-adapted salmonids, and this action could be independent of IGF production by the liver.

To our knowledge, only one study has been published on the effects of GH in tilapia (Clarke et al. 1977): in juvenile tilapia, high doses of homologous GH stimulated growth, as was indicated by increases in length and mass. We have shown before (Flik et al. 1985, 1986a,b) that, in tilapia, calcium uptake from the water via the gills is pivotal for growth and calcium homeostasis. The availability of recombinant tilapia GH (rtGH; Rentier-Delrue et al. 1989) allowed us to test homologous GH for its growth-promoting effect in freshwater tilapia. We focused our study on the effects of rtGH on calcium handling by the tilapia. Freshwater tilapia were weighed and injected with rtGH twice a week for up to 7 weeks. We measured the hepatosomatic index and the calcium balance of the fish and analyzed the plasma ion composition, the opercular ionocyte density and the bone and scale calcium and phosphate contents.

Tilapia, Oreochromis mossambicus Peters, were obtained from laboratory stock. The fish were held in 120-l all-glass aquaria with a continuous flow of Nijmegen city tapwater (0.7mmol l⁻¹ Ca²⁺, 0.2mmol l⁻¹ Mg²⁺, 0.5mmol l⁻¹ Na⁺, 0.06mmol l⁻¹ K⁺, pH7.8); the water temperature was 26±0.2°C. Lights were on for 12h per day. The loading density of fish did not exceed 1.5kg100 l⁻¹. Commercial trout pellets (Trouvit, Putten, The Netherlands) were administered by means of automated food dispensers throughout the light period at 2 or 5% of the recorded total mass of the groups of fish per day. Care was taken that all fish had access to the food and that all food was consumed. Experiments were carried out in May and June.

Recombinant tilapia growth hormone (rtGH; Rentier-Delrue et al. 1989) was dissolved in 0.02mol l⁻¹ NH₄HCO₃ (pH9.0) and diluted to the required concentration with phosphate-buffered saline (pH7.8). The dose was 100ng g⁻¹ fish. Injections were given intraperitoneally with a Hamilton precision syringe. The carrier volume was 1 μlg⁻¹ fish. Control animals received saline injections. Fish were injected twice a week, on Monday morning and Thursday afternoon. The fish were weighed before injection. The use of a Mettler PM34 Delta range balance allowed weighing of free-swimming fish and stress is therefore minimized compared with that seen with netting. Four groups of 15 fish each were included in an experiment: one group was weighed at the start and upon completion of the experiment, and not handled further (group U); one group was handled in the same way as the injected groups, but not injected (group H); one group served as injection controls (group S); and one group received rtGH injections (group GH). The holding conditions for all four groups were identical. The duration of the hormone treatment was 44 days.

Upon completion of the experiments, the fish were quickly anaesthetized in Tris-buffered (pH7.8) 3-aminobenzoic acid ethyl ester (MS-222; 1 gl⁻¹) and blood was collected (see below); subsequently, the fish were killed by spinal transection. The liver was removed and weighed to calculate the hepatosomatic index, defined as (W_l/W_f) ×100% (where W_l is the mass of the liver in grams and W_f is the mass of the fish in grams). One operculum was removed for determination of the ionocyte density. Vertebral bone and scale samples were collected to assess calcium and phosphate contents (see below).

Mixed arterial and venous blood was collected by puncture of the caudal vessels, using a heparinized tuberculin syringe fitted with a 23-gauge needle. Plasma was separated from cells by centrifugation (1min, 9000 g). Part of the plasma was ultrafiltered (Millipore Ultrafree-MC, molecular cut-off 10000Da) for the estimation of free calcium and magnesium levels. Total calcium and magnesium of plasma and ultrafiltered plasma were determined with commercial colorimetric kits (Sigma, St Louis, MO, USA). Combined calcium/phosphate and magnesium standards (Sigma) were used as reference. Plasma sodium and potassium contents were determined by flame emission spectrophotometry. Plasma osmolality was determined with a Roebling osmometer on fresh 50 μl plasma samples. Cortisol was determined on 5 μl of plasma by radioimmunoassay (Amerlex assay IM-2021, Amersham plc, UK), which is based on a highly specific antibody. The assay is not affected by the anticoagulant heparin. The recovery, defined as the increase in value when a known amount of cortisol is added to the sample and expressed as the percentage recovery (i.e. the measured increase divided by the predicted increase), varied between 98% and 103% in the range of the standard curve. The measurement range of this assay is 0–1700nmol l⁻¹. The sensitivity of the standard curve, estimated from the 95% confidence intervals of the within-assay variation of the zero standard counts, was 3±1nmol l⁻¹ (N=5). Serial dilutions of plasma samples yielded curves parallel to the standard curve.

Triplicate samples of 10 scales each were taken from both sides at the mid-lateral region, posterior to the operculum. A sample of vertebral bone was taken after removal of adhering muscle tissue by microwave cooking (1min, 700W). All tissues were weighed and dried to constant mass at 90°C, and the dry mass was determined to the nearest 0.01mg. Dried samples were dissolved in 0.5ml of concentrated HNO₃ at 60°C for 1h. The sample volume was brought to 5ml with double-distilled water. The total calcium was determined by a Thymol Blue method (for details, see Flik et al. 1986a) and phosphate was determined according to Fiske and Subbarow (1926).

Opercula were incubated for 1h in 2 μmol l⁻¹ 2-(dimethylaminostyryl)-1-ethylpyridinium iodine (DASPEI) dissolved in water. DASPEI stains the mitochondrion-rich chloride cells (ionocytes) (Bereiter-Hahn, 1976). After rinsing, the inner opercular epithelium was examined in a Zeiss fluorescence microscope at a magnification of 250×. Cells were counted in 20 different squares of the opercular epithelium with a total surface area of 5mm ² per fish.

The rtGh-treated fish and their controls were used 44–46 days after the start of hormone treatment. Unidirectional calcium influx was determined on the basis of initial uptake rates of ⁴⁵Ca²⁺ from the water and the water ⁴⁵Ca²⁺ specific activity, as described in detail elsewhere (Flik et al. 1989). Briefly, fish were exposed to water containing ⁴⁵Ca²⁺ of known specific activity. The fish were not fed during the flux determination. After 4h, the fish was anaesthetized, rinsed in 10mmol l⁻¹ CaCl₂ to remove tracer adsorbed to the body surface and quick-frozen on solid CO₂. Next, the fish was partly defrosted and the still frozen intestinal tract was removed. Subsequently, the ⁴⁵Ca²⁺ content of the fish and intestinal tract was determined after microwave-cooking and blending the cooked tissues with a known amount of distilled water in a Waring-type blender. Using this method, tracer uptake as a result of drinking is separated from branchial tracer uptake (Pang et al. 1980). The whole-body calcium influx determined in this way was shown to reflect extraintestinal branchial influx. Growing tilapia increase their total body calcium pool, and its size is directly related to the mass of the fish (see below). Therefore, net calcium influx may be determined on the basis of the growth-related increase in the total body calcium pool (the net influx equals the mean accumulation rate of calcium). The treatments reported here did not affect the calcium content of the bone, and therefore the total body calcium pool (Q_f) may be calculated as Q_f =357.5W_f^0.965μmol (W in g; Flik et al. 1985). Net calcium influx was calculated as ΔQ_f/Δt (time in h) over the 44 day period of the experiment. Calcium efflux (F_out) was calculated as the difference between branchial calcium influx (F_in) and net calcium influx (F_net), F_out=F_in-F_net.

The fish used in the experiments received food at a dose of 2 or 5% of their mass per day. To evaluate the effects of the amount of food provided on the calcium influx from the water, a comparison was made between the measured calcium influx and the predicted calcium influx at a food dose of 1% per day, given by the equation: F_in=50W_f^0.805 nmol h⁻¹ (W in g; Flik et al. 1985).

Values are presented as means ± S.D. Differences among groups were assessed by means of a one-way analysis of variance. Significance of differences between means was subsequently assessed by the Newman–Keuls test or the Mann–Whitney U-test, where appropriate. Linear regression analysis was based on the least-squares method. Significance was accepted when P<0.05.

rtGH treatment significantly stimulated growth as determined by the increase in body mass and length after 44 days (Table 1). These increases were significant when compared with those of unhandled (P<0.01) as well as with those of handled and saline-injected groups (P<0.05). No differences were observed among the control groups (U, H and S). The percentage increase in body mass over the initial mass after 44 days of treatment was 44.9% for the GH group and 25.6, 28.1 and 28.2% for groups U, H and S, respectively. The percentage increase in length was 15% for GH-treated fish and 8.4%, 9.1% and 9.1% for groups U, H and S, respectively.

Branchial calcium influx increased with the amount of food provided (Table 2). Male fish fed 5% per day had a calcium influx twice as high (P<0.05, Mann–Whitney U-test) as male fish fed 2%, 11.7 and 5.8 μmol h⁻¹ 100 g⁻¹, respectively. Calculated values for branchial calcium influx at a food dose of 1% are around 2 μmolh⁻¹ 100 g⁻¹. For male fish, the uptake of calcium from the water was positively correlated with the amount of food consumed according to: F_in=(2.36±0.18)X(% food)+0.11 μmol h⁻¹ 100 g⁻¹ (r=0.987; P<0.01). Female fish had a calcium influx significantly (P<0.01, Mann–Whitney U-test) lower than that of male fish [both fed 5%, 6.2 and 11.7 μmol h⁻¹ 100 g⁻¹, respectively]. The 5% food regimen represented ad libitum food conditions and was chosen for the GH experiment.

rtGH had no effects on calcium influx from the water (Table 3). As a result of the increased growth rate, rtGH-treated fish showed an enhanced net accumulation of calcium (9.45±0.60 μmol h⁻¹ 100 g⁻¹), 36% higher than in the untreated fish (P<0.025) and 25% higher than in the handled fish (P<0.05) and 37% higher than in the saline-injected fish (P<0.01). The calculated net efflux of calcium in rtGH-treated fish was, therefore, lower than in the controls.

Table 4 gives data on plasma ion composition and cortisol levels in fish treated for 44 days with rtGH. The data for the three control groups (U, H and S) have been pooled as no differences were observed among these groups for any of the variables measured. GH did not influence plasma ion composition but had a significant (P<0.025, Mann–Whitney U-test) effect on plasma cortisol levels, which were almost 50% lower than in the controls.

GH treatment increased the hepatosomatic index (HSI) compared with that of saline-injected (P<0.01) and handled and untreated fish (P<0.05) (Fig. 1). No differences were observed among groups U, H and S. In GH-treated fish, the HSI increased by 43 % compared with that of saline-treated fish (from 1.64 to 2.34%).

Analysis of vertebral bone and scales (Table 5) revealed no differences among the four groups of fish (U, H, S and GH) after 44 days of treatment. The values for the calcium content of vertebrae and scales are similar to previously reported values for tilapia (Flik et al. 1986b). The ratios of calcium to phosphate are in all cases close to the value of 1.43, representative for whitlockite, the probable form of apatite in fish bone (Herrmann-Erlee and Flik, 1989).

Opercular ionocyte densities increased in GH-treated fish compared with all other groups (P<0.01) (Fig. 2). The ionocyte density in the saline-injected group was significantly higher than in the unhandled (P<0.01) and handled (P<0.05) groups.

The results presented here demonstrate that tilapia recombinant growth hormone (Rentier-Delrue et al. 1989) is bioactive. The mild growth-promoting activity of rtGH in freshwater tilapia compares well with the mild effects of homologous recombinant GH in carp (Fine et al. 1993). The growth-stimulatory activity is further substantiated by its classical stimulatory effect on the hepatosomatic index. The growth-promoting effect of rtGH, therefore, appears to be mediated through stimulation of the liver, where the IGF is produced that subsequently enhances the growth of cartilage and bone. No change in the plasma ion composition was observed after rtGH treatment. The bone compartments increased in size but no changes in the mineral density occurred. Apparently, sufficient calcium, phosphate and bone matrix were available for calcium deposition to allow enhanced growth without influencing plasma ion composition or bone mineral density. The effect of GH on the calcium balance of tilapia is clearly different from that of prolactin (PRL). PRL does not stimulate growth, but stimulates calcium influx and reduces calcium efflux and, by doing so, induces hypercalcaemia and increases bone calcium density (Flik et al. 1986b; Swennen et al. 1991). We have reported before that low water levels of calcium (i.e. 0.2mmol l⁻¹ or less) may influence bone calcium content and plasma ion composition in tilapia (Flik et al. 1986a; Urasa and Wendelaar Bonga, 1987). Furthermore, it has been shown that tilapia mobilize calcium and phosphate from their bone, but only in times of shortage, e.g. during ovarian maturation and when diets are deficient in calcium (Urasa et al. 1985). It appears that fish do not show the strict calcium homeostasis observed in terrestrial vertebrates and rely primarily on external calcium (Perry and Flik, 1988). The observation that the unidirectional influx of calcium from the water is always larger than the calculated net influx (i.e. the growth-related calcium accumulation rate) into the fish suggests that fish rely on branchial calcium influx for growth and homeostasis: the amount of calcium offered in the food would have been sufficient to allow the growth observed (see below), yet the fish increased their branchial calcium influx. We realize that an estimation of the efficiency of intestinal calcium absorption may ultimately shed light on the role of intestinal calcium handling for growth.

A remarkable observation is that branchial calcium influx increased linearly with the amount of food offered to the fish. Given a food calcium content of 850 μmol g⁻¹ at an amount of 5% of its body mass per day, a 100g fish obtains 5×850=4250 μmol calcium. The Ca²⁺ influx via the gills per day is F_in×24=11.7×24=280.8 μmol for a 100g fish. Thus, although the food-associated calcium would suffice to cover the fish’s extra need for (enhanced) growth, a larger dietary calcium input does not lead to a decreased branchial calcium influx. Ample evidence has been given that calcium transport occurs in the proximal intestine of the tilapia (Flik et al. 1990). However, we have no information on the net movements of calcium in more distal segments of the intestine, where calcium secretion must occur. The role of dietary calcium in the calcium metabolism of the fish remains enigmatic: a calcium-deficient diet does not hamper calcium homeostasis or growth (Berg, 1970; Urasa and Wendelaar Bonga, 1987). Another line of evidence suggesting that branchial calcium uptake is adjusted to the growth rate of the fish comes from our data on female fish: under identical conditions, female fish have a lower calcium influx from the water, grow more slowly and do not reach the size of males (in our laboratory).

Recombinant tilapia GH did not enhance calcium influx (F_in) from the water. We realize that this statement is based on data that reflect a single time point, i.e. 44 days after the start of the treatment. It could well be that the unidirectional influx was stimulated earlier in the 44-day period. Such an increase could have accounted for an increased integrated net flux over the 44-day period, whereas the increased unidirectional influx had attenuated by day 44. The conclusion would be that rtGH transiently stimulated Ca²⁺ influx from the water to enhance calcium accumulation for growth. However, on the basis of the consistent stimulation of growth observed in this experiment, we consider this possibility to be unlikely. Moreover the Ca²⁺ influxes determined 16 days after the start of treatment on six saline-injected and five rtGH-injected fish were not significantly different (F_in=8.9±3.4 and 10.3±4.1 μmol h⁻¹ 100 g⁻¹ fish, for saline-and GH-treated fish, respectively; P>0.10), nor were these values different from those determined after 44 days of treatment. The Ca²⁺ influx in a fish provided with 5% food per day appears to be high enough to cover the growth-related accumulation of calcium. The rtGH-treated fish grew faster and enlarged their bone compartment (the bone calcium content did not change); these fish apparently accumulated more calcium per unit time (F_net increased). As a result, the calculated total efflux of calcium is lower in rtGH-treated fish. Direct measurements of branchial and extrabranchial calcium efflux are required to discriminate between branchial, renal and intestinal efflux routes as targets for rtGH.

The stimulation by rtGH of the opercular, and presumably the branchial, ionocyte density suggests an osmoregulatory action of the hormone. The total, DASPEI-stainable ionocyte population is not necessarily a good indicator of ion-transport activity of the gills. It may well be that the ionocyte turnover and number were stimulated by rtGH, but that at the same time the number of functional ionocytes, i.e. fully developed cells in contact with the water (Wendelaar Bonga et al. 1990), remained unchanged. However, we favour the idea that this rtGH effect in freshwater fish was an IGF-dependent mitogenic effect of exogenous GH. Clearly, these data require future evaluation of IGF levels. A stimulation of branchial ionocytes has been reported for seawater trout (Madsen, 1990); injection of GH into seawater fish improves hypo-osmoregulatory mechanisms and therefore the stimulation by GH of branchial ionocytes may be of functional significance.

The opercular ionocyte density in the control groups was somewhat higher than the densities reported before, 120–188mm⁻²versus less than 100mm⁻² in control freshwater tilapia (Wendelaar Bonga et al. 1990). We tentatively relate this increased ionocyte density to the relatively high cortisol levels (discussed below) seen in these fish (around 1000nmol l⁻¹). However, an increased cortisol level does not explain the increased ionocyte density in the GH-treated fish, as cortisol levels were about 50 % lower in these fish than in the controls. It would be interesting to know the dynamics of cortisol under these conditions. The lower level of cortisol could reflect an increased utilization of cortisol in rtGH-treated fish.

In general, the plasma cortisol levels in all four groups were high compared with those of completely unstressed fish (typical cortisol levels around 100nmol l⁻¹; G. Flik, unpublished observation). The present experimental set-up with frequent handling and an intensive feeding regimen, in all likelihood, leads to increased cortisol levels. Indeed the levels in groups U, H and S are elevated to a level normally seen in fish 10min after the start of netting. An alternative explanation for the lower cortisol levels in the GH group may be that the stress-related cortisol response in these fish has become suppressed. We did not observe differences between unhandled and handled control fish with respect to cortisol levels and, therefore, tentatively conclude that the elevated plasma cortisol levels somehow relate to the high dose of food provided for our fish. Again, this topic needs further research.

Lower cortisol levels in GH-treated fish may further indicate that GH exerts a negative control over the interrenals, a conclusion in line with the data of Carsia et al. (1985) on domestic fowl. It contrasts, however, with data on coho salmon by Young (1988), who reported a stimulatory action of (ovine) GH on the interrenals, both in vivo and in vitro. A negative feedback of GH on the interrenals, as suggested by our data, completes the functional interrelationship between cortisol and GH-producing cells proposed by Nishioka et al. (1985), who reported stimulatory effects of cortisol on GH release by tilapia pituitary gland in vitro.

The authors thank Tom Spanings for the excellent organization of the fish husbandry.