Igf1 and Igf2 stimulate growth and development of vertebrates. In mammals, liver-derived endocrine Igf1 mediates the growth promoting effects of GH during postnatal life, whereas Igf2 stimulates placental and fetal growth and is not regulated by GH. Insulin enhances Igf1 production by the mammalian liver directly, and by increasing hepatocyte sensitivity to GH. We examined the regulation of igf1 and igf2 mRNA levels by GH, insulin, and cortisol, and the effects of insulin and cortisol on GH sensitivity in primary cultured hepatocytes of tilapia, a cichlid teleost. GH increased mRNA levels of both igf1 and igf2 in a concentration-related and biphasic manner over the physiological range, with a greater effect on igf2 mRNA level. Insulin increased basal igf2 mRNA level, and strongly increased GH-stimulated igf2 mRNA level, but slightly reduced basal igf1 mRNA level and did not affect GH-stimulated igf1 mRNA level. Cortisol inhibited GH stimulation of igf1, but increased GH stimulation of igf2 mRNA level. The synergistic effect of insulin and GH on igf2 mRNA level was confirmed in vivo. These results indicate that insulin and cortisol differentially modulate the response of igf1 and igf2 mRNA to GH in tilapia hepatocytes, and suggest that the regulation of liver Igf2 production differs between fish and mammals. Regulation of liver Igf2 production in fish appears to be similar to regulation of liver Igf1 production in mammals.
The insulin-like growth factors (Igf1 and Igf2) are mitogenic peptides that regulate vertebrate growth (Humbel 1990, Jones & Clemmons 1995, Reinecke & Collet 1998). The Igfs are members of the evolutionarily ancient insulin-like family of peptides, found throughout the metazoans. Conserved features of the insulin-like peptides include regulation by nutritional status, and roles in nutrient metabolism, growth, development, reproduction, and aging (Tatar et al. 2003). Igf1 is thought to have arisen from insulin during the transition from chordates to primitive vertebrates, and Igf2 to have arisen from Igf1 in the common ancestor of the bony and cartilaginous fishes (Reinecke & Collet 1998). Both Igfs act through the same membrane receptor, the type 1 Igf receptor, and the activity of both Igfs is modulated by multiple Igf-binding proteins. Both Igfs have a dual mode of action as local and as endocrine growth factors. Local Igfs are produced throughout the body; whereas endocrine Igfs are mainly produced by the liver.
Liver-derived circulating Igf1 plays an essential role in postembryonic mammals as the primary mediator of GH-dependent growth. As developed in the somatomedin hypothesis, pituitary GH stimulates liver production of Igf1, which feeds back to inhibit GH secretion in the GH/Igf endocrine axis (Humbel 1990, Jones & Clemmons 1995, Phillips et al. 1998). In mice, both endocrine and local Igf1 play biologically significant roles in growth regulation (Stratikopoulos et al. 2008). Nutritional and metabolic status regulate the GH/Igf axis through direct effects on the liver, and by modulating liver sensitivity to GH. In catabolic states such as fasting and disease, liver production of Igf1 becomes resistant to stimulation by GH (Thissen et al. 1994, 1999). Metabolically responsive hormones link the anabolic/catabolic state of the animal to liver GH sensitivity. Insulin directly increases hepatocyte Igf1 production, and strongly increases hepatocyte responsiveness to GH in vivo (Griffen et al. 1987, Butler et al. 2003), and in mammalian and avian primary hepatocyte culture (Tollet et al. 1990, Boni-Schnetzler et al. 1991, Houston & O'Neill 1991, Phillips et al. 1998). Glucocorticoids induce liver GH resistance (Rodgers et al. 1994, Brameld et al. 1995, Beauloye et al. 1999).
In all vertebrate classes, igf2 is highly expressed during embryogenesis, and stimulates growth in embryonic tissues (Jones & Clemmons 1995, Reinecke & Collet 1998, Wood et al. 2005, White et al. 2009). In contrast, igf1 expression is low during embryogenesis and increases with the onset of postnatal GH-dependent growth. It is thought that liver Igf2 production is not strongly stimulated by GH or other hormones in postnatal mammals (Humbel 1990, Holly 1998). The lack of a clear picture of the regulation and function of endocrine Igf2 in postnatal mammals may be due to the central role of Igf2 in the development of the placenta. Igf2 is a principal regulator of the size and exchange capacity of the placenta. The parent–offspring evolutionary conflict theory predicts that this role as a mediator of maternal–fetal resource partitioning has resulted in selection on igf2 and divergence of Igf2 regulation and function (Haig 1993, O'Neill et al. 2007). A recent study in placental guppies supports this scenario (O'Neill et al. 2007). This suggests that Igf2 physiology may fundamentally differ between placental and non-placental vertebrates.
Recent studies show that teleost fishes including salmonids, sea bream, carp, catfish, eels, rabbitfish, sea bass, and hybrid striped bass, as well as the dogfish, an elasmobranch, differ from mammals in that liver igf2 gene expression and plasma Igf2 levels are stimulated by GH and respond to metabolic status (Shamblott et al. 1995, Tse et al. 2002, Vong et al. 2003a, Peterson et al. 2004, 2005, Carnevali et al. 2005, Gabillard et al. 2006, Wilkinson et al. 2006, Ayson et al. 2007, Terova et al. 2007, Gahr et al. 2008, Moriyama et al. 2008a, b, Picha et al. 2008b, Ponce et al. 2008, Devlin et al. 2009, Peterson & Waldbieser 2009, Eppler et al. 2010, Pierce et al. 2010). In primary cultured coho salmon hepatocytes, insulin directly increased igf2 mRNA, and strongly increased the response of igf2 mRNA to GH, but suppressed the response of igf1 mRNA to GH (Pierce et al. 2005, 2010). Thus, the regulation of liver igf2 gene expression in coho salmon is similar to the regulation of liver igf1 gene expression in mammals. We hypothesized that this may be the case in teleost fishes in general. To test this hypothesis, we examined the regulation of igf1 and igf2 gene expression by GH, insulin, and cortisol, and the effects of insulin and cortisol on GH sensitivity in primary cultured hepatocytes of a cichlid teleost, the Mozambique tilapia (Oreochromis mossambicus).
Materials and Methods
Mozambique tilapia were maintained in 700 l outdoor freshwater flow-through tanks at the Hawaii Institute of Marine Biology. Fish were fed ∼2% body weight per day in two daily feedings (Silver Cup Trout Chow, Portco Corporation, Vancouver, BC, Canada). Water temperature was maintained at 25–28 °C. Fish used in hepatocyte culture studies were 2- to 3-year-old males weighing 300–600 g, whereas fish used in the in vivo injection experiment were 1-year-old males weighing 60–80 g. All experiments were conducted under approved protocols in accordance with the principles and procedures of the Institutional Animal Care and Use Committee, University of Hawaii.
Hepatocyte isolation and culture
Fish were fasted for 1 day before hepatocyte isolation. Fish were first anesthetized by immersion in a nonlethal concentration of buffered tricaine methanesulfonate (MS-222, Argent Chemical Laboratories, Redmond, WA, USA, 0.5 g/l), and i.p. injected with lithium heparin (Sigma, 10 000 U/kg). After 5 min, fish were killed by immersion in a lethal concentration of MS-222 (2.0 g/l). Immediately after opercular movements stopped, fish were placed left side down on a sterile elevated screen. The right side of the peritoneal cavity was cut away. The gall bladder was punctured and drained, and the empty gall bladder and fat and connective tissue were dissected away, exposing the hepatic portal vein. The vessel was cut partway through and a polyethylene cannula was inserted and pushed toward the liver, until the tip was near where the vessel begins to ramify into the liver, and tied in place. The heart was removed to allow drainage. The liver was perfused with Ca++-free Hanks' buffered saline solution (HBSS) for 20–30 min at a flow rate of 1.5 ml/min. Perfusion then continued with 6.5–7.5 mg/100 ml collagenase (type IV, Sigma C5138, lot #085K8618), made fresh in Ca++-free HBSS and sterile filtered immediately before use. Collagenase perfusion continued until cracks appeared on the surface of the liver, typically 8–12 min. During Ca++-free HBSS and collagenase perfusion, the liver was massaged gently every 5 min. When digestion was complete, the liver was removed from the fish and placed in a sterile Petri dish in ice cold Ca++-free HBSS. Undigested sections of liver tissue, fat, connective tissue, and blood clots were dissected away, digested tissue was chopped fine with a razor blade, and cells were mashed through a 70 μm filter. Cells were collected in 50 μl of ice cold Ca++-free HBSS and pelleted (60 g, 5 min). Cells were resuspended in recovery medium (Ca++-free HBSS supplemented with 1.5 mM CaCl2, 2% cell culture grade BSA, 3 mM glucose, and 1× Gibco MEM essential and non-essential amino acid solutions). Cells were washed three times in recovery medium, and then settled on ice for 1–2 h. After settling, cells were resuspended in culture medium (modified RPMI 1640 medium, Gibco BRL; 20 mM HEPES, 5 mM NaHCO3, and 0.2% BSA) and washed twice. Cells were counted and examined for viability by trypan blue exclusion, and then plated at a density of 1–2×106 cells/ml, 0.5 ml/well on Falcon Primaria 24-well plates and incubated at 25 °C under plain air. Cells were isolated from individual fish and pooled if necessary. Cells were allowed to adhere to culture plates for an initial period as described in the Results section, and then medium was changed to test medium containing hormones. Native tilapia GH was purified by HPLC (Specker et al. 1985). Native bovine insulin and cortisol were purchased from Sigma (insulin Sigma I6634, lot 096K16911, activity 28 USP units/mg by HPLC). Ethanol (0.1%) was added to medium for all treatments in experiments testing the effects of cortisol. Cultures were stopped by adding Tri-Reagent (MRC, Cincinnati, OH, USA) to wells and freezing culture plates at −80 °C.
In vivo injection experiment
Freshwater-acclimated fish were fasted for 24 h and injected with hormones. Fish were not fed post-injection. During injection, fish were anesthetized (MS-222, 0.5 g/l) and injected with saline vehicle (0.9% NaCl, 1 μl/g body weight), native ovine GH (5 μg/g body weight, NIDDK oGH-15, lot AFP 9220A, activity 1.52 IU/mg in terms of Bovine GH International Standard 55/1), native bovine insulin (5 μg/g body weight, Sigma), or insulin and GH. Fish were fatally anesthetized 6 h later, and liver tissue was collected. Liver was homogenized immediately in Tri-Reagent and stored at −80 °C until RNA isolation.
RNA isolation, real-time quantitative RT-PCR, and plasma glucose assays
RNA was isolated following the MRC protocol, with bromochloropropane as the phase separation reagent and two 70% ethanol washes. RNA was quantified and purity assessed by spectrophotometry (Nanodrop ND-1000, Wilmington, DE, USA: RNA yield 286–906 ng/well, A260/280 1.90–2.18), RNA was diluted to 10–50 ng/μl, and first strand cDNA was synthesized with a kit (qScript cDNA Supermix, Quanta BioSciences, Gaithersburg, MD, USA). Hepatocyte and liver igf1 and reference gene (acidic ribosomal phosphoprotein P0 (arp) and elongation factor 1α (ef1α)) mRNA expression levels were quantified as described previously (Pierce et al. 2007, Breves et al. 2010). Hepatocyte and liver igf2 mRNA levels were quantified with a quantitative RT-PCR (qPCR) assay across the predicted exons 3/4 boundary (F primer: AGCCACCTCTCTACAGGTCATACC and R primer: ACTTCACGGTCACATGTTGCTT, hydrolysis probe: FAM-TGCCCGCACTAAAACAGGAAGTTCAGA-BHQ, amplicon size: 80 bp), which was validated (PCR efficiency 91–95%, biological template serial dilutions parallel to PCR product standard curve, no bias across PCR plate, no signal from genomic DNA or RT reactions with the reverse transcriptase omitted, single peak in melting curve analysis in assay run with SYBER green and probe omitted). All qPCR assays were run in 96-well format on an Applied BioSystems (ABI, Foster City, CA, USA) StepOne Plus qPCR machine, using standard cycling conditions. Assays were run using ABI Gene Expression Master Mix in a reaction volume of 15 μl, with primers and probe at a concentration of 200 nM. Gene expression levels for individual targets were quantified using PCR product standard curves and then normalized by dividing the expression level of a gene of interest in a given sample by the expression level of an appropriate reference gene in that sample as described previously (Pierce et al. 2007). Gene expression data for hepatocyte culture experiments were normalized to the expression level of arp, whereas gene expression data for the in vivo injection experiment were normalized to the expression level of ef1α. Statistically detectable differences in the CT for arp were not found between treatments in any hepatocyte culture experiment, and statistically detectable differences in the CT for ef1α were not found between treatments in the injection experiment, providing evidence that treatments did not affect our reference genes (ANOVA on replicate culture plate wells or fish). Plasma glucose was measured by the hexokinase method by a kit (GAHK-20, Sigma).
Gene expression data were log2 transformed before analysis. Treatment effects were examined by one- or two-way ANOVA, followed by the Student–Newman–Keuls test. Results were considered statistically detectable at P<0.05. Data analysis was conducted with Prism (GraphPad Software, San Diego, CA, USA). Data are given as mean±s.d. To establish tilapia hepatocyte culture, we used pooled cells (time course experiment: pooled cells from two fish and GH concentration-response experiment: pooled cells from two fish). In these experiments, our experimental unit is the well, and our statistical inferences are restricted to the pool of cells used. To show that effects were repeatable, we used independent cultures (insulin GH and factorial experiment: separately cultured cells from four fish and cortisol and GH factorial experiment: separately cultured cells from three fish). In these experiments, our experimental unit is the cell preparation, and our statistical inferences are to tilapia hepatocytes cultured using our methods. Owing to differences in the magnitude of response of different cultures, log2 transformed data from these experiments were normalized by multiplying by the average response to GH in all cultures divided by the response to GH in a given culture. To characterize concentration-response relationships, we used pooled cells (insulin concentration-response experiment: pooled cells from two fish and cortisol concentration-response experiment: pooled cells from two fish). In these experiments, our experimental unit is the well, and our statistical inferences are restricted to the pool of cells used.
Under our experimental conditions, freshly isolated tilapia hepatocytes adhered to culture plates after 4 h. Shorter adherence times resulted in substantial loss of cells when changing medium. Consequently, initial 4 h incubation in plain medium without hormones or serum was employed in all experiments. Gene expression levels of both igf1 and igf2 decreased rapidly during primary culture (Fig. 1, n=6–8 wells per point, two-way ANOVA igf1: time 55% of variation, P<0.0001; GH 30%, P<0.0001; interaction 12%, P<0.0001 and igf2: time 38%, P<0.0001; GH 30%, P<0.0001; interaction 26%, P<0.0001). After 24 h, control igf1 mRNA levels had decreased to 0.091±0.011 and igf2 mRNA levels to 0.128±0.009-fold time 0 levels. The addition of GH (100 ng/ml=4.5×10−9 M) to the medium increased mRNA levels of both igfs versus time-matched controls. Basal and GH-stimulated igf1 decreased with time in previous studies with primary tilapia and salmon hepatocytes (Schmid et al. 2000, Pierce et al. 2004). Based on the results for these and other transcripts, an incubation time of 6 h was selected, giving a total culture age of 10 h.
The response of both igfs–GH was concentration-dependent and biphasic (Fig. 2, n=7–12 wells per point, one-way ANOVA igf1: P<0.0001 and igf2: P<0.0001). For both igfs, mRNA levels were statistically detectably increased at 0.5×10−9 M GH, maximal response occurred at 5×10−9 M GH (igf1 1.92±0.12-fold 0 GH controls and igf2 8.36±0.54-fold 0 GH control), and the response was reduced to control levels at 500×10−9 M GH. Based on these results, co-incubation with 5×10−9 M GH was used to assess the effects of insulin and cortisol on GH sensitivity.
The effects of insulin (10−6 M) alone and in combination with GH (5×10−9 M) were tested in four primary hepatocyte cultures. Although the magnitude of effect varied, insulin, GH, and GH plus insulin treatment affected igf1 and igf2 mRNA levels similarly in all cultures (igf1: insulin: 0.89, 0.75, 0.80, 0.76; GH: 2.48, 1.94, 1.48, 1.48; GH plus insulin: 1.86, 1.68, 1.45, 1.63-fold controls respectively and igf2: insulin: 1.34, 3.27, 1.98, 4.03; GH: 1.82, 6.97, 3.16, 3.93; GH plus insulin: 2.79, 16.87, 7.63, 17.23-fold controls respectively). The effect of GH on igf1 and igf2 mRNA levels was statistically detectable in all cultures (one-way ANOVA on replicate culture wells). To present data from all cultures, log2 transformed data were normalized to the average level of stimulation of each igf by GH (Fig. 3, n=4 fish, one-way ANOVA igf1: P<0.0001 and igf2: P<0.0001). Insulin slightly suppressed basal igf1 mRNA levels (0.77±0.06-fold controls) but did not affect GH-stimulated igf1 mRNA levels. Insulin increased basal igf2 mRNA levels (2.44±0.42-fold controls), but to a lesser degree than GH (3.59-fold controls, calculation of s.d. precluded by normalization), and insulin plus GH strongly increased igf2 mRNA levels (2.71±0.46-fold GH alone).
The effect of insulin alone and in combination with GH (5×10−9 M) was further examined in a concentration-response study (Fig. 4, n=8 wells per point, two-way ANOVA igf1: insulin 1.8% of variation, P=0.0661; GH 73%, P<0.0001; interaction 2.9%, P=0.0043 and igf2: insulin 30% of variation, P<0.0001; GH 57%, P<0.0001; interaction 12%, P<0.0001). GH increased igf1 and igf2 mRNA levels at all insulin concentrations, including 0 insulin (igf1 1.73±1.6-fold controls and igf2 4.40±0.10-fold controls). Insulin did not affect the response of igf1 mRNA levels to GH, and did not decrease basal igf1 mRNA levels. An apparent slight increase in GH-stimulated igf1 mRNA level with increasing insulin was not statistically detectable. Insulin increased basal igf2 levels, and strongly increased the response of igf2 to GH. The increase in basal igf2 mRNA level became statistically detectable at 10−9 M insulin, and was maximal at 10−6 M insulin (3.20±0.43-fold 0 insulin controls). The increase in GH-stimulated igf2 mRNA level also became statistically detectable at 10−9 M insulin, and was maximal at 10−8 M insulin (2.83±0.18-fold GH alone).
The effects of cortisol (10−6 M) alone and in combination with GH (5×10−9 M) were tested in three primary hepatocyte cultures. Although the magnitude of effect varied, cortisol, GH, and GH plus cortisol treatment affected igf1 and igf2 mRNA levels similarly in all cultures (igf1: cortisol: 0.93, 1.11, 0.88; GH: 1.58, 1.52, 1.52; GH plus cortisol: 1.19, 1.31, 1.10-fold controls respectively and igf2: cortisol: 0.73, 0.77, 0.59; GH: 3.43, 4.35, 1.77; GH plus cortisol: 7.59, 8.85, 3.82-fold controls respectively). The effect of GH on igf1 and igf2 mRNA levels was statistically detectable in all cultures (one-way ANOVA on replicate culture wells). To present data from all cultures, log2 transformed data were normalized to the average level of stimulation of each igf by GH (Fig. 5, n=3 fish, one-way ANOVA igf1: P<0.0001 and igf2: P=0.0001). Cortisol did not statistically detectably affect basal igf1 mRNA levels, but suppressed GH-stimulated igf1 mRNA levels (0.78±0.038-fold GH alone). There was a consistent but not statistically detectable trend toward suppression of basal igf2 by cortisol (P=0.1312; 0.65±0.15-fold controls). Cortisol strongly increased GH-stimulated igf2 mRNA levels (2.65±0.81-fold GH alone).
The effect of cortisol alone and in combination with GH (5×10−9 M) was further examined in a concentration-response study (Fig. 6, n=7–8 wells per point, two-way ANOVA igf1: cortisol 13% of variation, P<0.0001; GH 48%, P<0.0001; interaction 4.5%, P<0.0001 and igf2: cortisol 2.2% of variation, P<0.0001; GH 84%, P<0.0001; interaction 5.8%, P<0.0001). GH increased igf1 and igf2 mRNA levels at all cortisol concentrations, including 0 cortisol (igf1 1.86±0.045-fold controls and igf2 3.74±0.089-fold controls). Cortisol did not affect basal igf1 mRNA levels, but reduced GH-stimulated igf1 mRNA levels. The decrease in igf1 mRNA level in cells co-incubated with GH became statistically detectable at 10−8 M cortisol and was maximal at 10−6 M cortisol (0.67±0.028-fold GH alone). Cortisol reduced basal igf2 mRNA levels at levels of 10−8 M and above (0.54±0.037-fold control levels at 10−6 M). In contrast, cortisol increased GH-stimulated igf2 mRNA levels. The increase became statistically detectable at 10−8 M cortisol, was maximal at 10−7 M (1.50±0.065-fold GH alone), and was statistically detectably reduced from maximal levels at 10−6 M.
In an analysis of all hepatocyte culture experiments employing 5×10−9 M GH, GH stimulation of igf2 mRNA levels was statistically detectably greater than GH stimulation of igf1 mRNA levels (igf1 1.7±0.10-fold control; igf2 4.2±0.65-fold control; P=0.0017, t-test, n=10).
Fish were injected with insulin and GH to confirm the results of our in vitro experiments (Fig. 7, n=7–8 fish per point, one-way ANOVA glucose: P=0.0010; igf1: P<0.0001; and igf2: P=0.0002). Insulin and GH doses similar to those used in previous studies in fish were used, and plasma glucose was measured to confirm the expected hypo- and hyperglycemic effects of insulin and GH respectively (Mommsen & Plisetskaya 1991, Wood et al. 2005). Insulin injection (bovine, 5 μg/g body weight) reduced plasma glucose levels at 6 h. GH injection (ovine, 5 μg/g body weight) caused apparent increase in plasma glucose. However, this increase was not statistically detectable under our experimental conditions (α=0.05; probability of type 1 error=0.156 and probability of type 2 error=0.220). Nevertheless, co-injection of GH with insulin prevented the insulin-induced decrease in plasma glucose. GH injection increased liver mRNA levels of both igfs (igf1: 1.94±0.17 and igf2: 7.08±0.75-fold controls). An apparent decrease in igf1 mRNA level with insulin injection was not statistically detectable, and insulin injection did not change igf2 mRNA level; however, co-injection with GH and insulin increased mRNA levels of both igf1 and igf2 compared with levels produced by injection with GH alone (igf1: 1.43±0.12 and igf2: 1.52±0.13-fold GH alone). The effects of GH and GH plus insulin on igf2 mRNA levels were statistically detectably greater than effects on igf1 mRNA levels.
This study shows that igf1 and igf2 mRNA levels are differentially regulated by insulin and cortisol in tilapia hepatocytes through direct effects and differential modulation of GH sensitivity. Insulin increased basal igf2 mRNA levels and strongly increased the response of igf2 mRNA levels to GH, but did not increase basal igf1 mRNA levels or the response of igf1mRNA levels to GH. Similar results were found in salmon hepatocytes (Pierce et al. 2005, 2010), suggesting that insulin and GH synergistically stimulate liver Igf2 but not Igf1 production in teleost fishes.
Both igf1 and igf2 mRNA levels were stimulated by GH in tilapia primary hepatocytes, as has been found in previous in vitro studies in teleost fishes and elasmobranches (Shamblott et al. 1995, Schmid et al. 2000, Pierce et al. 2004, 2010, Leung et al. 2008, Moriyama et al. 2008a, b). Recent in vivo studies have shown that GH treatment increases liver igf2 mRNA levels in a wide variety of fish species (see Introduction for citations), which is consistent with our results. Circulating GH levels in tilapia range from ∼0.05 to 1×10−9 M (Weber & Grau 1999, Uchida et al. 2003, Fox et al. 2009), suggesting that physiological regulation of liver igf1 and igf2 by GH occurs on the left side of the concentration-response curve. In humans, GH injection increased liver Igf2 transcripts from promoters P2 and P4, but did not change overall Igf2 transcript levels, and GH treatment increased P2 transcripts in primary human hepatocytes (Olivecrona et al. 1999, von Horn et al. 2002). As far as we are aware, this is the only recent report of GH-dependent liver Igf2 gene expression in mammals. This suggests that teleost fishes differ from mammals in that fish liver igf2 gene expression is GH-dependent.
Incubation with insulin stimulated igf2 mRNA level in tilapia primary hepatocytes, and co-incubation with insulin and GH strongly increased igf2 mRNA levels over levels with insulin or GH alone. In contrast, basal igf1 mRNA level was slightly suppressed by insulin, and GH-stimulated igf1 mRNA level did not respond to insulin. A concentration-response study showed strong enhancement of the stimulation of igf2 from 10−10 to 10−8 M insulin in the presence of GH. In the same cells, igf1 responded reliably to GH, but the response was not modulated by insulin, suggesting that the synergistic interaction between insulin and GH in the regulation of igf2 mRNA level is specific to this transcript. Although circulating insulin levels in tilapia have not been measured, fish plasma insulin ranges from 0.2 to 5×10−9 M (Mommsen & Plisetskaya 1991), suggesting that regulation of igf2 occurs over the physiological range. In salmon hepatocytes, basal and GH-stimulated igf2 responded similarly to insulin, whereas GH-stimulated igf1 was suppressed by insulin (Pierce et al. 2005, 2010). Thus, insulin stimulation of basal liver igf2 but not igf1 gene expression, and enhancement of the response of liver igf2 but not igf1 to GH, appear to be conserved in teleost fishes.
Injection of hormones confirmed in vivo GH stimulation of liver igf2 gene expression, and a positive interaction between insulin and GH in stimulation of liver igf2 gene expression. However, insulin injection alone did not result in a statistically detectable increase in liver igf2 gene expression, which differs from our in vitro results. This may have been caused by the decrease in blood glucose seen after insulin injection, or a cortisol response to the injection. Glucose positively regulates igf2 gene expression in fetal rat hepatocytes (Goya et al. 1999), and in this study, cortisol tended to suppress basal igf2 in vitro. Injection of GH-stimulated igf1 gene expression, consistent with our in vitro results and previous studies (Kajimura et al. 2001, Eppler et al. 2010). Injection of insulin tended to decrease igf1 gene expression (not statistically detectable), consistent with the slight suppression of basal igf1 gene expression by insulin seen in vitro. However, co-injection of GH and insulin increased igf1 gene expression versus GH alone, which differs from our in vitro results. This may have been an indirect effect of insulin. In brown trout, insulin injection increased plasma Igf1 3 h later (Banos et al. 1999), even though in coho salmon hepatocytes, insulin did not affect basal igf1 and suppressed GH-stimulated igf1 (Pierce et al. 2005), suggesting that indirect stimulatory effects of insulin on liver Igf1 production may exist in salmonids. In this study, GH-stimulated hepatic igf2 mRNA levels to a biologically significantly greater degree than igf1 mRNA levels in vitro and in vivo, consistent with recent results in GH transgenic tilapia (Eppler et al. 2010). This suggests that liver-derived endocrine Igf2 may play a larger role than liver-derived endocrine Igf1 in mediating the growth-stimulating effects of GH in tilapia.
Insulin plays a central role in the regulation of liver Igf1 production in mammals. Insulin secretion into the hepatic portal vessels results in exposure of the liver to high concentrations of insulin. Maintenance of insulin levels is necessary to maintain liver GH sensitivity, liver igf1 gene expression, blood Igf1 levels, and growth (Griffen et al. 1987, Rodgers et al. 1994, Thissen et al. 1994, Phillips et al. 1998, Butler et al. 2003). Insulin levels in the portal vessels are also elevated in fish (Plisetskaya & Sullivan 1989). In mammalian and avian primary hepatocyte culture, insulin increases basal, and GH-dependent Igf1 mRNA level and protein secretion (Tollet et al. 1990, Boni-Schnetzler et al. 1991, Houston & O'Neill 1991, Denver & Nicoll 1994, Phillips et al. 1998). The results of the present and our previous studies suggest that the regulatory linkage between the pancreatic islets and the liver may also operate in fish, in the regulation of liver Igf2 production.
The differences found between mammals and fish in the regulation of liver igf mRNA levels by GH and insulin implies divergence of gene regulation. In mammals, current models suggest that GH stimulation of liver Igf1 transcription is mediated by binding of the transcription factor Stat5b to multiple enhancer sites dispersed across at least 135 kb throughout the Igf1 locus (Rosenfeld & Hwa 2009, Chia et al. 2010). In fish, GH can stimulate liver igf1 transcription via Stat5 and the liver-specific transcription factor Hnf-1α (Meton et al. 1999, Vong et al. 2003b), suggesting that this pathway may be conserved. In salmonids, GH stimulation of igf2 gene expression has been proposed to be mediated by transcription factors in the CCAAT/enhancer-binding protein (C/EBP) family (Shamblott et al. 1998, Palamarchuk et al. 2001). In rainbow trout, GH treatment increased liver igf2 gene expression but reduced liver C/EBPβ2 occupancy of binding sites dispersed over at least 5 kb in the igf2 locus, suggesting that the role of C/EBPβ2 may be negative (Lo et al. 2007, Lo & Chen 2010). However, in carp, an igf2 promoter region sufficient for response to GH did not contain binding sites for C/EBP, Stat, or Hnf family transcription factors (Tse et al. 2008). The discrepancies between studies may be due to the operation of multiple long-range enhancers. Further studies are required to determine the transcription factors and gene regulatory elements that mediate GH stimulation of igf2 transcription in tilapia. In rats, an igf1 promoter positive response element that interacts with a novel insulin-response element-binding protein (IRE-BP1) downstream of PI3-kinase has been identified (Villafuerte et al. 2004). Insulin regulation of fish igf2 promoters has not been investigated.
In the presence of GH, cortisol reduced igf1 and increased igf2 mRNA levels in tilapia hepatocytes. Modulation of GH response occurred at 10−8–10−6 M cortisol, coinciding with the physiological range from basal to stressed in tilapia (Balm et al. 1994, Breves et al. 2010). Cortisol injection reduced liver igf1 gene expression and circulating Igf1 in vivo in tilapia (Kajimura et al. 2003). In sea bream hepatocytes, cortisol reduced basal igf1 mRNA levels at physiological concentrations (Leung et al. 2008), and in coho salmon hepatocytes, the glucocorticoid receptor agonist dexamethasone strongly inhibited GH-stimulated igf1 mRNA levels (Pierce et al. 2005). Thus, glucocorticoid suppression of igf1 expression, either directly or via GH resistance, appears to be conserved in teleost fishes. Glucocorticoids reduce liver IGF1 production in mammals (Rodgers et al. 1994, Brameld et al. 1995, Beauloye et al. 1999). This study and our previous study in coho salmon hepatocytes show glucocorticoid stimulation of igf2 mRNA levels, either directly or via enhanced GH response (Pierce et al. 2010). Therefore, the most reasonable hypothesis at this point is that glucocorticoids inversely regulate liver production of Igf1 and Igf2 in fishes. This suggests that liver-derived endocrine Igf1 and Igf2 may have different functions in fishes. It is possible that the two Igfs differentially regulate organ and tissue growth. Cortisol increases in fishes during life history transitions in which tissue remodeling and differential growth occur, such as salinity adaptation, smoltification in salmonids, and reproductive maturation.
This and other studies provide evidence that liver-derived endocrine Igf2 may be a somatomedin in tilapia and other teleost fishes. Studies on the function of Igf2 in fish support this idea. In tilapia, administration of either recombinant native Igf1- or Igf2-stimulated growth (Chen et al. 2000). Circulating Igf2 is regulated by metabolic status in salmonids (Wilkinson et al. 2006). Both Igfs stimulated mitogenic endpoints in zebrafish embryos (Pozios et al. 2001), and rainbow trout muscle cell culture (Codina et al. 2008). Both Igfs exert negative feedback on pituitary GH secretion (Fruchtman et al. 2000). If endocrine Igf2 is a somatomedin in fishes, then both Igfs should be taken into account in applications of growth endocrinology in fisheries management, conservation, and aquaculture. Plasma levels of both Igfs may be useful indicators of growth status in wild and captive fishes (Picha et al. 2008a). Plasma levels of both Igfs may provide information on energetic status to neuroendocrine systems that regulate major life history transitions, such as smolting and reproductive maturation (Dickhoff et al. 1997, Luckenbach et al. 2010). Clearly, conclusions regarding Igf2 physiology based on experiments in mammalian model systems cannot be extended to teleost fishes (White et al. 2009). Further study of the physiology of the two Igfs from a comparative perspective will clarify the functional evolution of the GH/Igf system (O'Neill et al. 2007), and may provide an answer to ‘the Igf2 enigma’ (Humbel 1990, Holly 1998).
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
This project was supported by National Research Initiative Competitive Grants nos 2006-35206-16647 and 2008-35206-18785 from the USDA National Institute of Food and Agriculture and by National Science Foundation Grants nos IOB05-17769 and OISE08-52518.
We thank the members of the Grau lab for assistance: Dr Darren Lerner, Dr Kai Fox, Dr Lori Davis, Nancy Visitacion, Eli Witt, Anna Kosztowny, Masa Yoshioka, and Ryan Helms. We thank the anonymous referees and statistical reviewer whose comments improved this paper.
BeauloyeVKetelslegersJMMoreauBThissenJP1999Dexamethasone inhibits both growth hormone (GH)-induction of insulin-like growth factor-I (IGF-I) mRNA and GH receptor (GHR) mRNA levels in rat primary cultured hepatocytes. Growth Hormone and IGF Research9205–211. doi:10.1054/ghir.1999.0110.
ButlerSTMarrALPeltonSHRadcliffRPLucyMCButlerWR2003Insulin restores GH responsiveness during lactation-induced negative energy balance in dairy cattle: effects on expression of IGF-I and GH receptor 1A. Journal of Endocrinology176205–217. doi:10.1677/joe.0.1760205.
CodinaMGarcia de la serranaDSanchez-GurmachesJMontserratNChistyakovaONavarroIGutierrezJ2008Metabolic and mitogenic effects of IGF-II in rainbow trout (Oncorhynchus mykiss) myocytes in culture and the role of IGF-II in the PI3K/Akt and MAPK signalling pathways. General and Comparative Endocrinology157116–124. doi:10.1016/j.ygcen.2008.04.009.
FoxBKBrevesJPHiranoTGrauEG2009Effects of short- and long-term fasting on plasma and stomach ghrelin, and the growth hormone/insulin-like growth factor I axis in the tilapia, Oreochromis mossambicus. Domestic Animal Endocrinology371–11. doi:10.1016/j.domaniend.2009.01.001.
GriffenSCRussellSMKatzLSNicollCS1987Insulin exerts metabolic and growth-promoting effects by a direct action on the liver in vivo: clarification of the functional significance of the portal vascular link between the beta cells of the pancreatic islets and the liver. PNAS847300–7304. doi:10.1073/pnas.84.20.7300.
KajimuraSUchidaKYadaTRileyLGByattJCCollierRJAidaKHiranoTGrauEG2001Stimulation of insulin-like growth factor-I production by recombinant bovine growth hormone in Mozambique tilapia, Oreochromis mossambicus. Fish Physiology and Biochemistry25221–230. doi:10.1023/A:1022268811599.
LoJHChiouPPLinCMChenTT2007Molecular cloning and expression analysis of rainbow trout (Oncorhynchus mykiss) CCAAT/enhancer binding protein genes and their responses to induction by GH in vitro and in vivo. Journal of Endocrinology194393–406. doi:10.1677/JOE-07-0085.
MoriyamaSYamaguchiKTakasawaTChibaHKawauchiH2008bIdentification of two insulin-like growth factor IIs in the Japanese eel, Anguilla japonica: cloning, tissue distribution, and expression after growth hormone treatment and seawater acclimation. Comparative Biochemistry and Physiology149B47–57..
OlivecronaHHildingAEkstromCBarleHNybergBMollerCDelhantyPJBaxterRCAngelinBEkstromTJ1999Acute and short-term effects of growth hormone on insulin-like growth factors and their binding proteins: serum levels and hepatic messenger ribonucleic acid responses in humans. Journal of Clinical Endocrinology and Metabolism84553–560. doi:10.1210/jc.84.2.553.
PichaMETuranoMJBeckmanBRBorskiRJ2008aEndocrine biomarkers of growth and applications to aquaculture: a minireview of growth hormone, insulin-like growth factor (IGF)-I, and IGF-binding proteins as potential growth indicators in fish. North American Journal of Aquaculture70196–211. doi:10.1577/A07-038.1.
PierceALDickeyJTLarsenDAFukadaHSwansonPDickhoffWW2004A quantitative real-time RT-PCR assay for salmon IGF-I mRNA, and its application in the study of GH regulation of IGF-I gene expression in primary culture of salmon hepatocytes. General and Comparative Endocrinology135401–411. doi:10.1016/j.ygcen.2003.10.010.
PierceALFoxBKDavisLKVisitacionNKitahashiTHiranoTGrauEG2007Prolactin receptor, growth hormone receptor, and putative somatolactin receptor in Mozambique tilapia: tissue specific expression and differential regulation by salinity and fasting. General and Comparative Endocrinology15431–40. doi:10.1016/j.ygcen.2007.06.023.
PierceALDickeyJTFelliLSwansonPDickhoffWW2010Metabolic hormones regulate basal and growth hormone-dependent igf2 mRNA level in primary cultured coho salmon hepatocytes: effects of insulin, glucagon, dexamethasone, and triiodothyronine. Journal of Endocrinology204331–339. doi:10.1677/JOE-09-0338.
PonceMInfanteCFunesVManchadoM2008Molecular characterization and gene expression analysis of insulin-like growth factors I and II in the redbanded seabream, Pagrus auriga: transcriptional regulation by growth hormone. Comparative Biochemistry and Physiology150B418–426. doi:10.1016/j.cbpb.2008.04.013.
ShamblottMJLeungSGreeneMWChenTT1998Characterization of a teleost insulin-like growth factor II (IGF-II) gene: evidence for promoter CCAAT/enhancer-binding protein (C/EBP) sites, and the presence of hepatic C/EBP. Molecular Marine Biology and Biotechnology7181–190..
TerovaGRimoldiSChiniVGornatiRBernardiniGSarogliaM2007Cloning and expression analysis of insulin-like growth factor I and II in liver and muscle of sea bass (Dicentrarchus labrax, L.) during long-term fasting and refeeding. Journal of Fish Biology70219–233. doi:10.1111/j.1095-8649.2007.01402.x.
TolletPEnbergBModeA1990Growth hormone (GH) regulation of cytochrome P-450IIC12, insulin-like growth factor-I (IGF-I), and GH receptor messenger RNA expression in primary rat hepatocytes: a hormonal interplay with insulin, IGF-I, and thyroid hormone. Molecular Endocrinology41934–1942. doi:10.1210/mend-4-12-1934.
WeberGMGrauEG1999Changes in serum concentrations and pituitary content of the two prolactins and growth hormone during the reproductive cycle in female tilapia, Oreochromis mossambicus, compared with changes during fasting. Comparative Biochemistry and Physiology124323–335. doi:10.1016/S0742-8413(99)00081-X.
WilkinsonRJPorterMWoolcottHLonglandRCarragherJF2006Effects of aquaculture related stressors and nutritional restriction on circulating growth factors (GH, IGF-I and IGF-II) in Atlantic salmon and rainbow trout. Comparative Biochemistry and Physiology145214–224. doi:10.1016/j.cbpa.2006.06.010.